In Vivo Monitoring of Multiple Trace Metals in the Brain Extracellular

In Vivo Monitoring of Multiple Trace Metals in the Brain Extracellular Fluid of ... trace metals in the extracellular fluid (ECF) in the brains of ane...
0 downloads 0 Views 396KB Size
Anal. Chem. 2007, 79, 8900-8910

In Vivo Monitoring of Multiple Trace Metals in the Brain Extracellular Fluid of Anesthetized Rats by Microdialysis-Membrane Desalter-ICPMS Y. T. Chung,† Y. C. Ling,† C. S. Yang,§ Y. C. Sun,*,‡ P. L. Lee,‡ C. Y. Lin,‡ C. C. Hong,‡ and M. H. Yang‡

Department of Chemistry, and Department of Biomedical Engineering and Environmental Sciences, National Tsing-Hua University, Hsinchu, Taiwan, and Department of Applied Chemistry, National Chi-Nan University, Nantou, Taiwan

We have developed an on-line analytical system involving microdialysis (MD) sampling, a carbohydrate membrane desalter (CMD), and an inductively coupled plasma mass spectrometer (ICPMS) system for the simultaneous determination of multiple trace metals in the extracellular fluid (ECF) in the brains of anesthetized rats. The microdialysate that perfused from the animal at a flow rate of 0.5 µL/min was on-line transferred to the CMD to remove the high-sodium matrix, followed by ICPMS measurement. The role of the CMD in this on-line system was investigated in detail. With prior addition of EDTA to the microdialysate to form anionic complexes of the metal analytes and the use of NH4Cl as a regenerant to exchange Na+ with NH4+ ions, both quantitative recovery of the trace metal analytes and quantitative removal of the sodium matrix could be achieved. Two experimental modes of the monitoring system were constructed. For those metals (e.g., Cu, Zn, and Mn) that existed at (sub)nanogram-permilliliter concentrations in the microdialysate, the temporal resolution was 10 min when using a 10 µL loop for sample collection, followed by CMD and ICPMS; for those elements (e.g., Ca and Mg) that existed at microgram-permilliliter levels (or greater), near-real-time analysis was possible because the microdialysate could be led, bypassing the sample loop, directly to the CMD for desalting without any time delay. Further improvement of the temporal resolution for the low-concentration elements was not possible without decreasing the detection limits of mass detection. Among the eight trace metals tested using this on-line system, the method detection limits for Cu, Zn, Mn, Co, Ni, and Pb reached subnanogram-permilliliter levels; for electrolyte species such as Ca and Mg, the detection limits were in the range of 50-100 ng/mL. Analytical accuracy, expressed as spike recovery, was 100% ( 15% for all of the elements tested. We demonstrate the applicability of the proposed system through the successful measurement of the basal values of Ca, Mg, Cu, Zn, and Mn in the ECF of a living rat brain and through in vivo monitoring of the concentration profiles of Mn and Pt in the ECF after the injection of drugs (MnCl2 and cisplatin) into the rats. This microdialysis system is 8900 Analytical Chemistry, Vol. 79, No. 23, December 1, 2007

the first to offer real-time, in vivo monitoring of trace elements such as Ca and Mg. Metals play important roles in metabolism, biosynthesis, and many other biochemical activities, for example, iron in oxygen transportation,1 manganese in the synthesis of glutamine synthase,2-4 and zinc as a component of neurotransmitters in the central nervous system (CNS).5-9 In addition to their vital roles, the presence of metals is detrimental under certain circumstances; for example, the deposition of iron in the brain can initiate a series of pathophysiological progressions of chronic brain diseases.10-12 It is generally believed that the vital and detrimental roles of metals in the body are concentration- and tissue-dependent. Dietary intake is the major route through which metals enter the body; primarily, they appear in the blood as a result of digestive absorption. The circulating metals would have to extravasate from the blood vessels and enter the extracellular fluids (ECFs) of target organs before they could further enter the cells. There have been reports indicating the possibility for chronic interorgan transportation of metals; for this process to occur, the intracellular metals must reversely exit the cells and enter into the ECFs before being transported further. Therefore, profiling the time-dependent concentrations of metals in the organ ECFs would provide important information regarding the possible * To whom correspondence should be addressed. Fax: +886-3-5723883. Phone: +886-3-5715131, ext 35596. E-mail: [email protected]. † Department of Chemistry, National Tsing-Hua University. ‡ Department of Biomedical Engineering and Environmental Sciences, National Tsing-Hua University. § National Chi-Nan University. (1) Kendrick, M. J.; May, M. T.; Plishka, M. J.; Robinson, K. D. Metals in Biological Systems; Ellis Horwood Limited: Great Britain, 1992. (2) Aschner, M.; Aschner, J. L. Neurosci. Biobehav. Rev. 1991, 15, 333-340. (3) Aschner, M. Environ. Health Perspect. 2000, 108 (Suppl. 3), 429-432. (4) Tiffany-Castiglioni, E.; Qian, Y. Neurotoxicology 2001, 22, 577-592. (5) Assaf, S. Y.; Chung, S. H. Nature 1984, 308, 734-736. (6) Westbrook, G. L.; Mayer, M. L. Nature 1987, 328, 640-643. (7) Koh, J. Y.; Suh, S. W.; Gwag, B. J.; He, Y. Y.; Hsu, C. Y.; Choi, D. W. Science 1996, 272, 1013-1016. (8) Frederickson, C. J.; Koh, J. Y.; Bush, A. I. Nat. Rev. Neurosci. 2005, 6, 449-462. (9) Sarkar, B. Biological Aspects of Metals and Metal-Related Diseases; Raven Press: New York, 1983. (10) Lovell, M. A.; Robertson, J. D.; Teesdale, W. J.; Campbell, J. L.; Markesbery, W. R. J. Neurol. Sci. 1998, 158, 47-52. (11) Andrasi, E.; Farkas, E.; Gawlik, D.; Rosick, U.; Bratter, P. J. Alzheimer’s Dis. 2000, 2, 17-26. (12) Bush, A. I. Trends Neurosci. 2003, 26, 207-214. 10.1021/ac070981z CCC: $37.00

© 2007 American Chemical Society Published on Web 11/01/2007

trafficking of metals in the body and lead to a greater understanding of the adsorption, metabolism, and excretion processes of biological metals. The brain, which is the central organ of the CNS, is compartmentalized by specialized barrier systems to prevent it from the direct assault of metals. The ECF of the brain comprises cerebrospinal fluid (CSF) and fluids extravasated from capillaries; the distribution of metals within this ECF is seriously regulated and might correlate to the neurotrophic or neurodegenerative state of brain. To obtain additional information regarding the roles metals play in physiological behavior, it would be very helpful if we could devise analytical methods for quantifying the concentrations of several analytes simultaneously and recording their kinetic changes faithfully in vivo. Presently, there are a number of novel in vivo methods that allow us to glance into the brain, such as magnetic resonance imaging (MRI), positron emission tomography (PET), and direct probe insertion techniques. Although invasive, probes such as biosensor, voltammetry, and microdialysis (MD) systems could be inserted to gather information directly, dynamically, and specifically on a site of interest. The in vivo voltammetry technique is highly sensitive and nearly real-time responsive, but it suffers from difficulties when analyzing much more complex systems in which different compounds might oxidize at the same potential difference.13-15 In another approach, by setting the environment isotonic carefully across a dialysis membrane, MD systems could be used to extract a target analyte according to its concentration gradient and reflect its status in the ECF.15-19 No biological fluid is removed from the milieu during this sampling process, and the whole biological system is only minimally disturbed.16 MD systems do, however, possess certain characteristics that might hinder their use in continuous detection applications. For MD sampling, the perfusate must be prepared at a high degree of salinity to mimic a real biological fluid; thus, saline solution or Ringer solution is usually utilized. The detector might, therefore, suffer to a high degree from a matrix effect, and pretreatmentsto remove the salinitysis always required. Moreover, because a flow rate of a few microliters per minute (typically 0.2-3.0 µL/min) is commonly adopted in MD sampling, only tiny amounts of sample (usually less than 10 µL) are employed, which would decrease the performance of the measurement system. For successful in vivo monitoring of metals in microdialysates, an ideal detection system would exhibit high durability toward a salty matrix as well as high sensitivity. Among the most highly sensitive spectrometric methods, electrothermal atomic absorption spectrometry (ETAAS) with an appropriate temperature program and matrix modifier could eliminate most of the matrix, inductively coupled plasma mass spectrometry (ICPMS) equipped with a reaction cell20,21 or mixed-gas plasma22 would lessen the effect of (13) Parkin, M. C.; Hopwood, S. E.; Strong, A. J.; Boutelle, M. G. TrAC, Trends Anal. Chem. 2003, 22, 487-497. (14) Stamford, J. A.; Justice, J. B. Anal. Chem. 1996, 68, 359A-363A. (15) de Lange, E. C. M.; de Boer, A. G.; Breimer, D. D. Adv. Drug Delivery Rev. 1999, 36, 211-227. (16) Lunte, C. E.; Scott, D. O.; Kissinger, P. T. Anal. Chem. 1991, 63, 773A780A. (17) de Lange, E. C. M.; de Boer, A. G.; Breimer, D. D. Adv. Drug Delivery Rev. 2000, 45, 125-148. (18) Davies, M. I.; Cooper, J. D.; Desmond, S. S.; Lunte, C. E.; Lunte, S. M. Adv. Drug Delivery Rev. 2000, 45, 169-188. (19) Boschi, G.; Scherrmann. J. M. Adv. Drug Delivery Rev. 2000, 45, 271-281.

polyatomic interference, and high-resolution mass spectrometry (HR-MS) could provide the same features, with even higher sensitivity and lower interference, but at a higher cost.23,24 Although use of these systems would remove most of the spectral interference, an alternative calibration strategy would be required to alleviate nonspectral interference.22 To develop an analytical technique for in vivo monitoring of multiple elements in the ECF of a living rat brain, the use of MD on-line coupling in conjunction with ICPMS measurement appears to be suitable method of choice. Although ICPMS provides a multielement analysis capability with high detection sensitivity, the presence of a high content of sodium ions in the sample might create a serious matrix effect that would deteriorate the analytical results. There are a number of documented methods for removing the sodium matrix from complex mixtures, such as biological and seawater samples; the most common are solvent extraction, knotted reactor (KR),25-29 and flow injection methods using columns packed with PTFE beads30 or other types of resin.31-35 Solvent extraction is hard to couple on-line with the MD sampling probe and ICPMS, and thus, it is ruled out. The tiny sample amounts and lower recoveries render KR as a less-than-useful technique. Column-based methods retaining either the analyte or matrix inside the column would do well to discriminate the analyte from the salty matrix, but the important ability to concentrate the analyte from the samples would be lost because only a tiny sample size (often smaller than 10 µL) would be available by microdialysis. In addition to the techniques described above, the permselective membrane techniquesso-called membrane desaltingsbased on the exchange of cations (H3O+ in most cases) with Na+ on the other side of the membrane can be used preferentially as a means of desalting. The device has been used many times in conjunction with IC-ICPMS for successfully separating analytes from Na+ matrix ions. A membrane desalter composed of a pair of carbohydratebased cation-exchange membranes can be used to efficiently and nonspecifically desalt all of the cations that flux inside. Therefore, to achieve effective removal of the sodium matrix while preserving the analytes of interest in the dialysate sample, in this study we (20) Leonhard, P.; Pepelnik, R.; Prange, A.; Yamada, N.; Yamada, T. J. Anal. At. Spectrom. 2002, 17, 189-196. (21) Louie, H.; Wu, M.; Di, P.; Snitch, P.; Chapple, G. J. Anal. At. Spectrom. 2002, 17, 587-591. (22) Holliday, A. E.; Beauchemin, D. J. Anal. At. Spectrom. 2003, 18, 11091112. (23) Rodushkin, I.; Ruth, T.; Klockare, D. J. Anal. At. Spectrom. 1998, 13, 159166. (24) Field, M. P.; Cullen, J. T.; Sherrel, R. M. J. Anal. At. Spectrom. 1999, 14, 1425-1431. (25) Benkhedda, K.; Infante, H. G.; Ivanova, E.; Adams, F. C. J. Anal. At. Spectrom. 2000, 15, 1349-1356. (26) Chen, H. H.; Beauchemin, D. J. Anal. At. Spectrom. 2001, 16, 1356-1363. (27) Salonia, J. A.; Wuilloud, R. G.; Ga´squez, J. A.; Olsina, R. A.; Martinez, L. D. J. Anal. At. Spectrom. 1999, 14, 1239-1243. (28) Yang, X. P.; Jiang, Y. TrAC, Trends Anal. Chem. 2001, 20, 552-562. (29) Wang, J.; Hansen, E. H. J. Anal. At. Spectrom. 2002, 17, 1278-1283. (30) Wang, J.; Hansen, E. H. J. Anal. At. Spectrom. 2002, 17, 248-252. (31) Lee, K. H.; Oghima, M.; Motomizu, S. Analyst 2002, 127, 769-774. (32) Lin, P. H.; Danadurai, K. S. K.; Huang, S. D. J. Anal. At. Spectrom. 2001, 16, 409-412. (33) Berman, S. S.; Willie, S. N.; Desaulniers, J. A. H. Anal. Chem. 1981, 53, 2337-2340. (34) McLaren, J. W.; Mykytuik, A. P.; Willie, S. N.; Berman, S. S. Anal. Chem. 1985, 57, 2907-2911. (35) Kumagai, H.; Yamanaka, M.; Sakai, T.; Yokoyama, T.; Suzuki, T. M.; Suzuki, T. J. Anal. At. Spectrom. 1998, 13, 579-582.

Analytical Chemistry, Vol. 79, No. 23, December 1, 2007

8901

employed a modified membrane separation system. The metal analytes in the microdialysate were converted into anionic complexes through reactions with EDTA prior to leading them into the desalting unit. Meanwhile, to avoid dissociation of the metal-EDTA complexes as a result of a change in the pH during the desalting process, an ammonium chloride solution (at pH 8), rather than the more conventional dilute sulfuric acid solution, was used as the regenerant solution to maintain the pH of the sample solution within a suitable range throughout the separation process. To date, there have been two studies of off-line ETAAS36,37 coupled with MD and several reports of on-line MD coupled with flame AAS (FAAS)38,39 and ETAAS40-43 to monitor the dynamic changes of Ca and Mg in the blood of rabbits and Mn, Zn, and Mg in the brain of an anesthetized rat. In these studies, the salty matrixes were previously eliminated by adjusting the temperature program and choosing a suitable matrix modifier, such as ammonium nitrate. Temporal resolutions of 25 and 50 min were reported for Mn and As, respectively. These methods could be useful, however, to determine the change of an individual element in a microenvironment of an animal, but they could not be applied to simultaneous multielement detection. Recently, we reported the coupling of MD with on-line in-tube solid-phase extraction and ICPMS for in vivo monitoring of trace elements.44 In this study, we continuously monitored the dynamic changes of Cu, Zn, and Mn ions in the ECF of the brain in vivo, but the temporal resolution achievable was limited to only 15 min. In the present study, we attempted to explore the feasibility of shortening the temporal resolutionshopefully near to real timesfor the simultaneous analysis of multiple elements in the ECF of the brains of living rats. To meet this purpose, we employed the unique features of a membrane suppressor as a desalting interface system for this on-line analytical system. The membrane desalter system, unlike other desalting techniques such as KR and column-based methods that are founded on the separation principles of adsorption and desorption of the analyte, provides the possibility of continuous and real-time separation of analytes from salty matrixes. This situation arises because the separation using the membrane suppressor is based on a countercurrent exchange mechanism (H3O+ exchange for Na+) involving a regenerant (normally H2SO4) and because the microdialysate can flow continuously through the desalting system without a time delay. It is, therefore, possible that the microdialysate from in vivo sampling can bypass the sample loop used for collecting the tiny flow volume of the microdialysate and be lead directly to the (36) Itoh, T.; Saito, T.; Fujimura, M.; Watanabe, S.; Saito, K. Brain Res. 1993, 618, 318-322. (37) Itoh, T.; Saito, T.; Watanabe, S.; Saito, K. Trace Elem. Electrolytes 1996, 13, 196-199. (38) Tseng, W. C.; Sun, Y. C.; Lee, C. F.; Yang, M. H.; Huang, Y. L. Anal. Sci. 2005, 21, 225-229. (39) Tseng, W. C.; Sun, Y. C.; Lee, C. F.; Chen, B. H.; Yang, M. H.; Huang, Y. L. Talanta 2005, 66, 740-745. (40) Tseng, W. C.; Sun, Y. C.; Yang, M. H.; Chen, T. P.; Lin, T. H.; Huang, Y. L. J. Anal. At. Spectrom. 2003, 18, 38-43. (41) Yang, D. Y.; Lee, J. B.; Lin, M. C.; Huang, Y. L.; Liu, H. W.; Liang, Y. J.; Cheng, F. C. J. Am. Coll. Nutr. 2004, 23, 552S-555S. (42) Lee, M. S.; Wu, Y. S.; Yang, D. Y.; Lee, J. B.; Cheng, F. C. Clin. Chim. Acta 2002, 318, 121-125. (43) Lin, M. C.; Huang, Y. L.; Liu, H. W.; Yang, D. Y.; Lee, C. P.; Yang, L. L.; Cheng, F. C. J. Am. Coll. Nutr. 2004, 23, 561S-565S. (44) Sun, Y. C.; Lu, Y. W.; Chung, Y. T. J. Anal. At. Spectrom. 2007, 22, 77-83.

8902

Analytical Chemistry, Vol. 79, No. 23, December 1, 2007

membrane desalter and, subsequently, the ICPMS for measurement, provided that the analyte concentration in the microdialysate is sufficiently high. In this present study, we aimed to develop an on-line analytical system that could be used for the simultaneous determination of multiple trace elements in the brain ECF of a living animal. Two modes of monitoring were constructed for this technique. For those elements (e.g., transition metals) that exist in the very low concentrations (subnanograms per milliliter) in the ECF, the microdialysate was treated through prior collection in a sample loop over a fixed time interval to provide a sufficient amount of sample that would meet the sensitivity requirements for ICPMS measurement; for those elements (e.g., Ca and Mg) that exist in relatively higher concentrations (micrograms per milliliter) in the ECF, the microdialysate bypassed the sample loop and proceeded directly to the membrane desalter and subsequent ICPMS measurement. In the former case, the temporal resolution shortened to ca. 10 min; in the latter case, the monitoring could be achieved at close to real time. The performance of the on-line system was evaluated with respect to the accuracy, precision, and long-term stability. The applicability of our proposed system was tested for its utility in the measurement of the basal values of various trace metals in the ECF of a rat brain and in the in vivo monitoring of the concentration profiles of Mn and Pt in the ECF after injection of drugs (MnCl2 and cisplatin) into the living rat. EXPERIMENTAL SECTION Reagents and Containers. Deionized water was obtained after purification through deionization and double distillation (Milli-Q Element, Millipore, Bedford, MA). The stock saline solution as perfusate was prepared by dissolving an appropriate amount of ultrapure grade sodium chloride (Merck, Darmstadt, Germany) in deionized water; it was purified using a self-packed Chelex-100 column. To mimic normal physiological conditions, the pH of the saline solution was adjusted to ca. 7.4 using 1 M NH4OH and/or 1 M HNO3. High-purity EDTA (>99.5%) was purchased from Fluka (Switzerland). Stock solutions (1000 mg/ L) of the analytes Ca, Mg, Cu, Zn, Mn, Co, Ni, and Pb were purchased from E. Merck (Darmstadt, Germany); the working aqueous standards were prepared afresh by further dilution with saline solution. The 600 mM NH4Cl, which served as the regenerant solution in the desalter, was prepared by mixing equal amounts of 10% (v/v) NH4OH and HCl. The solution of MnCl2 used in the in vivo experiment was prepared in the saline solution at physiological pH; cisplatin was purchased from Bristol-Myers Squibb (Sermoneta, Italy) as a ready-to-use solution having a concentration of 100 mg cisplatin/L. Polytetrafluoroethylene (PTFE), perfluoroalkoxy (PFA), and glass containers were cleaned through immersion in 6 M HNO3 overnight. The tubing used to connect the components was perfused with high-purity water for at least 2 h before each experiment. To avoid metal contamination, a fully plastic syringe from Norm-Ject (Henke Sass Wolf GMBH, Tufflingen, Germany) was used to deliver the perfusate and the EDTA diluent. Apparatus and Instrumentation. Figure 1 provides a depiction of the MD-desalter-ICPMS hyphenation system, which consists of three main parts: the MD probe, the membrane desalter, and the ICPMS. The MD sampling system consists of a microinjection syringe pump (Cole Parmer) and a 24 mm long

Figure 1. Schematic illustration of the on-line MD-desalter-ICPMS hyphenation system for continuous monitoring of diffusible analytes in the ECF of the brain of anesthetized rat (modes a and b): P1 and P2, syringe pumps; P3 and P4, peristaltic pumps; P5, piston pump; MDP, microdialysis probe; W, waste; IV, injection valve; SL, sample loop; D, diluent; CS, carrier solution; RS, regenerant solution.

MD probe (CMA/20, CMA, Solna, Sweden) with a 4 mm long, 0.5 mm diameter, metal-free polycarbonate (PC) membrane that had a molecular weight cutoff (MWCO) of 20 kDa. Connection of the microinjection syringe pump (P1) to the inlet of the MD probe and of the outlet of the MD probe to the micro-Tee (Alltech Associates, Inc., Deerfield, IL) was accomplished using fluorinated ethylene polypropylene (FEP) tubing (internal volume, 1.2 µL/ 100 mm length) purchased from CMA. Two modes of monitoring were constructed, as depicted in Figure 1. In mode a, the dialysate was mixed on-line with EDTA that perfused by another microinjection syringe pump (P2), and the mixture was collected in a 10 µL sample loop (SL) that was mounted on a autoinjection valve (IV). After collecting for a period of time (10 min) the sample was delivered to the desalter by a carrier solution (CS; 10 mg EDTA/L) at a flow rate of 100 µL/ min, pumped by a piston pump (P5, Rheos 2000, Flux Instruments, Switzerland). In mode b, no sample loop was used in the system; the dialysate was mixed on-line with EDTA (D) to evaluate the possibility of real-time analysis. Nevertheless, a peristaltic pump (P3, MAB 7, Alltech) was adopted to conquer the backpressure inside the desalter (reported by the manufacturer to be ca. 100 psi). Moreover, the CS was used either to reduce the residence time of the sample inside the desalter or to make the flow rate compatible with the nebulizer equipped on the ICPMS. A carbohydrate membrane desalter (CMD, Dionex Corp., Sunnyvale,) equipped with a pair of cation-exchange membranes was used to separate Na+ ions from the microdialysate. Samples that flowed into the central channel were desalted by exchanging Na+ ions with H3O+ in the regenerant solution outside the membrane. Electrolysissapplying a constant current on the desalterswas used to further enhance and accelerate the removal efficiency of Na+ ions. NH4Cl solution (600 mM)srather than the dilute (7 mM) H2SO4 solution recommended by the suppliers was used as the regenerant solution. To provide a suitable

membrane potential to generate water electrolysis in the CMD, the minimum current was set at 50 mA. The ICPMS used was a Micromass Platform (Micromass Ltd., Manchester, U.K.) equipped with a hexapole collision cell; helium and hydrogen were chosen as buffer and reaction gases, respectively, to remove any possible interference of argides. A PFA nebulizer (PFA-50, Cetac Technologies, Omaha, NE) suitable for analysis of tiny samples was installed; it was fitted to a standard Scotty double-pass spray chamber, with temperature control provided by a water-cooled chiller at 4 °C. The instrumental operating conditions selected for optimal sensitivity and low background noise are presented in Table 1. Prior to each day’s experiments, minor fine-tuning was performed to the ion lens and reaction cell settings, e.g., the cone voltage (CV), hexapole bias (HB), and hydrogen and helium gas flow rates, with tuning solution that contained 10 µg/L each of Mn, Cu, Zn, and Pb. In Vivo Experiment. Adult male Sprague-Dawley (SD, 350450 g) and Wistar (350 g) rats were obtained from the Laboratory Animal Center at the National Science Council of the Republic of China (Taipei, Taiwan). These animals were acclimated to their environmentally controlled quarters (25 °C and 12:12 h lightdark cycle) for at least 1 week before experimentation and fed ad libitum with a standard diet and water. The rats were treated under the regulations of the “Principles of laboratory animal care” (NIH publication no. 86-23, revised, 1985); the animal experiments were performed under the approval of the Committee of Experimental Animals of National Tsing-Hua University. The rats were fasted overnight and were then deeply anesthetized by urethane (1200 mg/kg body weight, intraperitoneally) which efficacy could remain throughout the whole experimental period (>12 h). After mounting the head of the rat on a stereotaxic apparatus (Davis Kopf Instruments, Tujunga, CA), a midline incision of the skull was executed. The MD probe that was perfused with saline solution at a flow rate of 0.5 µL/min was Analytical Chemistry, Vol. 79, No. 23, December 1, 2007

8903

Table 1. Operating Conditions for the MD-Desalter-ICPMS System Microdialysis Sampling 4 mm × 0.5 mm membrane, metal-free and MWCO at 20 kDa perfusion solution saline (0.9% NaCl) solution (pH 7.4) perfusion flow rate 0.5 µL/min sample loop 10 µL probe (CMA 20)

Desalter carrier solution carrier flow rate regenerant solution regenerant flow rate

10 mg EDTA/L 100 µL/min 600 mM NH4Cl 7 mL/min ICPMS

ICPMS spectrometer Ar flow rates plasma auxiliary nebulizer plasma forward power sampling cone skimmer cone cone voltage collision cell hexapole bias H2 flow rate He flow rate

Micromass Platform 13.7 L/min 0.90 L/min 1.04 L/min 1350 W nickel, 1.14 mm orifice nickel, 0.89 mm orifice 125 V -0.8 V 3.0 L/min 5.0 L/min

implanted into the exposed brain (5 mm anterior, 5 mm lateral to the bregma, and 5 mm from the brain surface). After 2 to 3 h (to reach equilibrium), the basal values of the analytes were monitored continuously, while the concentration was quantified through an in vitro calibration curve that was set up before animal testing. After finishing the experiment, the MD probe was removed from the brain and preserved in deionized water. Probe recovery was used only as an index to evaluate the integrity of the membrane; in most cases it could be confined within the range of 40-50%. The MD probe was discarded if its probe recovery was lower than 30%. Application of the system to in vivo monitoring of Mn and Pt was conducted after intraperitoneal administration of MnCl2 (100 mg/kg rat body) or cisplatin (4 mg/kg rat body weight) into anesthetized rats. Following the administration of drugs, the concentration changes of the analytes were monitored for 400 min. RESULTS AND DISCUSSION Microdialysis sampling of ECF has emerged as an important technique in biomedical, pharmaceutical, and neuroscience applications. Coupling an MD probe to ICPMS provides a useful and powerful tool for obtaining basal levels and dynamic information simultaneously for multiple metal analytes in living organisms. There are, however, two important issues that should be taken into account prior to using such MD-based monitoring systems for the measurement of trace element concentrations in ECFs: the detection sensitivity and long-term stability of the system for prolonged monitoring. Because extremely low basal concentrations of metal analytes are present in ECF [normally in (sub)nanogram-per-milliliter levels] and high contents of salt are present in the dialysates, it is of primary importance to control the analytical blank to achieve the lowest possible detection limits 8904

Analytical Chemistry, Vol. 79, No. 23, December 1, 2007

and to remove as much salt as possible from the dialysate to prevent matrix loading on the ICPMS. In toxico- and pharmacokinetic studies using in vivo microdialysis, dynamic changes of the analytes must be monitored for several hours; therefore, longterm stability of the on-line system is necessary. The factors influencing the stability of this on-line analytical system mainly arise from changes in the performance of the MD probe after long-term operation and also from nonspectral interference caused mainly by deposition of the matrix on the sampling cone or ion lenses of the ICPMS measurement system. This factor can be evaluated or eliminated through a recovery test of the MD probe and desalting of the dialysate prior to ICPMS measurement. The on-line analytical system depicted in Figure 1 has three main parts: the MD, desalter, and ICPMS components. Because MD and ICPMS are both well-established techniques for in vivo sampling and multielement determination, the main emphasis of this study was the development of suitable on-line desalting and high-temporal-resolution analytical procedures and their application in the in vivo monitoring of trace metals in the brains of living rats. Microdialysis Sampling. Because microdialysis is a continuous sampling method, normally the dialysate is collected over a fixed time interval to provide a sufficient sample volume, with each sample representing an average concentration obtained over this time interval. The low concentrations (usually in the nanogramper-milliliter regime) of metal analytes such as Cu, Zn, and Mn in ECF means that the temporal resolution is usually determined by the time required to remove detectable quantities of these analytes. To estimate the in vivo concentrations of metals, it is necessary to determine the recovery of the MD probe and the linearity of the on-line analytical system. In this study, a commercial metalfree probe (CMA-20) incorporating a relatively short (4 mm) MD membrane was used because of considerations regarding the depth of a rat’s brain. It is well-known that the extraction efficiency of an analyte will decrease upon increasing the perfusion flow rate; thus, to ensure the performance of the system, it was important to balance the perfusion flow rate, sample volume, and concentration detection limits. In this study, we perfused the saline solution at a flow rate of 0.5 µL/min through the MD probe; the dialysate, after mixing with EDTA solution supplied at an equal flow rate of 0.5 µL/min, was subsequently loaded into a 10 µL sample loop of the on-line injector and further transferred to the ICPMS for measurement. The linearity of the system was within the range from 1 to 40 ng/mL for Cu, Zn, Mn, Co, Ni, and Pb and from 1 to 40 µg/mL for Ca and Mg. The recoveries of all of the elements tested with the present MD system were in the range of 40-50%. We found that the recovery of the MD probe did not change appreciably after continuous operation for up to 12 h, indicating the feasibility of using the present MD sampling system for prolonged monitoring. Optimization of the Membrane Desalter. The Dionex CMD used in this study was originally designed to remove Na+ ions from sodium acetate effluent after anion-exchange chromatography with pulsed amperometric detection (HPAEC/PAD) oligosaccharide separations. The CMD employs a countercurrent cationexchange mechanism (H3O+ exchanged for Na+) driven by both electrolytically generated and pneumatically supplied regenerant

Figure 2. Effect of the concentration of NH4Cl on the pH change of the effluent and the removal efficiency of Na+ ions: sample solution, 0.9% NaCl; sample flow rate, 2 mL/min; pH of NH4Cl solution, 8; electrolytic current, 50 mA.

(H2SO4 or trifluoroacetic acid are used as the source for pneumatically supplied H3O+). In this study, the membrane desalter was, however, used not only to remove high contents of Na+ ions but also to preserve metal analytes in the microdialysate. Separation of Na+ Ions. The removal of Na+ ions utilizing permselective membrane techniques, such as membrane suppressor45-48 or membrane desalter45-50 approaches, is based on the exchange of H3O+, which is provided by a regenerant solution (usually H2SO4) and water electrolysis of the anode, with Na+ ions across the cation-exchange membranes through mass action and an applied electric field. The membrane desalter can, however, remove all of the cations efficiently, but not specifically. To preserve the cations of interest in the separation process, EDTA can be used to transform the metal analytes into stable anionic complexes. When the combination ratio of EDTA to metal ions is 1:1, the metal complexes are negatively charged and cannot be exchanged for H3O+ by the membrane suppressor. Because the stability of metal complexes is pH-dependent, to maintain a satisfactory degree of association between EDTA and most transition metal and alkaline earth metal ions would required values of pH of g7 and g11, respectively. According to a preliminary study, we found that when the sample solution in 0.9% NaCl was passed through the CMD, in which 1% (v/v) H2SO4 was used as regenerant solution, the pH of the effluent dropped remarkably from its original value of 7 to (45) Gu ¨ rleyu ¨ k, H.; Wallschla¨ger, D. J. Anal. At. Spectrom. 2001, 16, 926-930. (46) Garcia-Ferna´ndez, R.; Garcia-Alonso, J. I.; Sanz-Medel, A. J. Anal. At. Spectrom. 2001, 16, 1035-1039. (47) Garcia-Ferna´ndez, R.; Garcia-Alonso, J. I.; Sanz-Medel, A. J. Chromatogr., A 2004, 1033, 127-133. (48) Garcia-Ferna´ndez, R.; Garcia-Alonso, J. I.; Sanz-Medel, A. J. Anal. At. Spectrom. 2004, 19, 649-651. (49) Thayer, J. R.; Rohrer, J. S.; Avdalovic, N.; Gearing, R. P. Anal. Biochem. 1998, 256, 207-216. (50) Cataldi, T. R. I.; Campa, C.; De Benedetto, G. E. Fresenius’ J. Anal. Chem. 2000, 368, 739-758.

less than 2. Thus, the metal ions of interest would not be preserved in the sample solution as a result of dissociation of the anion complexes at such a low pH. Consequently, H2SO4 cannot be used as the regenerant; instead, we investigated the use of NH4Cl to exchange NH4+ for Na+ ions. The applicability of NH4Cl as a regenerant for the removal of Na+ ions from salty matrixes was investigated using the established on-line system. The experiments were undertaken by passing the saline solution (pH 7) at a flow rate of 2 mL/min through the CMD and subsequently to the inductively coupled plasma optical emission spectrometer (ICPOES) for the detection of residual Na. Figure 2 displays the effect of the concentration of the NH4Cl solution on both the pH change of the effluent and the removal efficiency of Na+ ions. The results indicate that the pH of the effluent reduced to 2.5-3.5 when lower concentrations (from 50 to ca. 400 mM) of NH4Cl (pH 8) were applied, but then it increased abruptly (to pH 7) when the concentration of NH4Cl was >450 mM. We attribute the reduction of the pH in the lower concentration range of NH4Cl to the diffusion of H3O+, generated by the electrolysis of water, through the membrane; as the concentration of NH4Cl was increased, however, an overwhelming amount of NH4+ ions passed through the membrane, resulting in the pH increasing to its original level (pH 7). We also note from Figure 2 that the maximum removal of Na+ ions under the present experimental conditions was only ca. 70%. Further improvement of the desalting efficiency was attempted using NH4Cl as the regenerant solution at a selected concentration of 600 mM in the following study. To enhance the removal efficiency of Na+ ions, we investigated the effects of the flow rates of the sample and the regenerant. Figure 3 displays the composite effect on Na+ ion removal of variations of the two flow rates. When the flow rate of the sample was reduced to 0.1 mL/min, quantitative removal of Na+ ions was achieved regardless the flow rate of the regenerant, which we varied from 4 to 20 mL/min. We attribute this finding to the fact Analytical Chemistry, Vol. 79, No. 23, December 1, 2007

8905

Figure 3. Effect of the flow rates of the sample and regenerant on the Na+ removal efficiency: sample solution, 0.9% NaCl; concentration of NH4Cl solution, 600 mM; pH of NH4Cl solution, 8; electrolytic current, 50 mA.

that, at a lower sample flow rate, Na+ ions will have sufficient time to diffuse through the membrane suppressor to exchange with NH4+ in the regenerant. Figure 3 also indicates that the removal efficiency of Na+ ions apparently decreased upon increasing the flow rate of sample (to 0.5 or 1.0 mL/min, as indicated). Nevertheless, this unwanted result could be overcome effectively when utilizing a higher flow rate of the regenerant solution. In terms of real sample analysis using this on-line system, the flow rate of the microdialysate entering the desalter would be less than 0.1 mL/min; consequently, we selected flow rates of 0.1 and 7 mL/min for the sample and regenerant solutions, respectively, for subsequent studies. When using saline solutions as the eluent (Na content, ca. 3500 mg/L), the residual concentration of Na+ in the effluent was determined to be less than 0.5 mg/L, indicating the near-quantitative removal efficiency of this desalting system. Recovery of Analyte Ions. Similar to other ion-exchange membrane separations, metal analytes that exist as cations would also be removed as Na+ ions from the sample to the regenerant in this CMD system. This situation requires the mandatory application of EDTA to convert the analyte ions into anionic forms. Figure 4 illustrates the effect of changing the EDTA-to-analyte molar ratios on the recovery of analytes in the effluent. The recoveries of the tested analytes generally increased upon increasing the molar ratio of EDTA to metal ions; quantitative levels were achieved when the molar ratio exceeded a value of 5. The behavior of the change in recovery was, however, different from element to element, as typified by the cases of Cu and Mn. This phenomenon can be explained by considering the fact that Cu possesses a higher formation constant (log Kf ) 18.8) and can form stable complexes at relatively lower EDTA concentrations; the lower formation constant for Mn (log Kf ) 13.6) would require a higher EDTA concentration. On the basis of these results, a large excess of EDTA would be required (molar ratio g5) to ensure complete formation of all metal complexes. To be on the safe side, in this study we employed a molar ratio of ca. 10, which is equivalent to ca. 10 µg EDTA/mL, based on an estimated analyte concentration of 10 8906 Analytical Chemistry, Vol. 79, No. 23, December 1, 2007

Figure 4. Effect of the EDTA-to-analyte molar ratio on the recovery of analytes: total analyte concentration, 10 µg/L; sample solution, 0.9% NaCl; concentration of NH4Cl buffer, 600 mM; pH of NH4Cl solution, 8; electrolytic current, 50 mA.

µg/L in the microdialysate sample. Our experimental results revealed that EDTA at this concentration level would not cause any detrimental effects on the nebulizer or the vacuum interface (clogging of the cones) in the ICPMS apparatus. Analytical Performance of the On-Line System. On the basis of the experimental results obtained above for microdialysis and membrane desalting, we attempted to optimize the on-line system. As depicted in Figure 1, the microdialysate from in vivo sampling was allowed to proceed through two different pathways to the next processes of desalting and ICPMS measurement. For elements such as transition metals that exist at very low levels [(sub)nanograms per milliliter] in the ECF, the microdialysate must be treated through collection in a sample loop over a fixed time interval to provide a suitable sample that would meet the sensitivity requirements for ICPMS measurement; for elements such as Ca and Mg that exist in relatively higher concentrations (micrograms per milliliter) in ECF, the microdialysate can bypass the sample loop and enter the membrane desalter directly and then proceed further to the ICPMS for measurement. Because the separation event in the membrane suppressor is based on countercurrent exchange of NH4+ for Na+ supplied by the NH4Cl regenerant solution, the microdialysate can flow continuously through the desalting system without any time delay. This situation could, therefore, provide the unique possibility of allowing continuous monitoring of changes in analyte concentrations (e.g., for Ca and Mg) in close to real time. We evaluated the applicability of this MD-desalter-ICPMS system for the simultaneous determination of multiple elements in microdialysate samples in terms of the signal interference of the system and its long-term stability. We expected that the ability to reduce the Na+ ion content to negligible levels in microdialysate samples after separation by the membrane suppressor would alleviate isobaric interference caused by polyatomic ions and prevent sodium deposition on the sampling cone, thus improving the long-term stability of the ICPMS signals. Our experimental

Table 2. Analytical Characteristics of the Established MD-Desalter-ICPMS System element

MDL, ng/mL

precision, % (RSD)a

spike recovery, %a

Mn Co Ni Cu Zn Pb Ca Mg

0.44 0.69 0.97 0.70 0.31 0.78 25 112

6 4 8 10 4 7 0.66 0.87

99 97 90 88 99 100 b b

a

Number of experiments (n ) 3). b Data not available.

results clearly indicated that polyatomic interference on the measurement of some analyte ions, e.g., for 23Na40Ar, 23Na35Cl, and 23Na16O16O by 63Cu, 58Ni, and 55Mn, respectively, was eliminated; in addition, depression of the sensitivity toward some typical elements, such as Cu, Zn, and Mn, caused by shifting of plasma equilibrium and salt buildup on the cone tip, was alleviated significantly. Thus, use of the membrane desalter did indeed improve the instrumental stability and analytical performance. Prior to in vivo application, we evaluated the analytical performance of the on-line system in terms of its detection sensitivity, precision, accuracy, and long-term stability. In view of the extremely low basal concentration of most metal ions, except for Ca and Mg, in the ECF of rat brain and the limited amount of sample collected through microdialysis sampling, it was vitally important to have strict control over the blank value. In this study, all of the reagents used, including the saline solution (NaCl), complexing agent (EDTA), and regenerant solution (NH4Cl), were of ultrahigh-purity grade. The blank was determined by following the established process of this on-line system, using the saline solution as the blank sample. The method detection limits (MDLs) were estimated based on 3 times the standard deviation of the signal of the blank sample collected with a 10 µL sample loop (n ) 7); Table 2 lists the results. The values of MDL for the eight elements tested were in the (sub)nanogram-per-milliliter level. According to literature reports, the basal values of Cu, Zn, and Mn in the ECF of rat brain lie in the range between 1 and 40 ng/mL, with Ca and Mg present at microgram-per-milliliter levels. Thus, our system should be suitable for in vivo monitoring of, at least, this set of elements. Because no certified reference material having a matrix similar to the ECF was available, we examined the accuracy of our proposed on-line method by using a saline solution spiked with known concentrations of the analytes (15 µg/mL for each tested element). The spike recovery for all of the elements tested was within the range of 95-105% (n ) 3). The system stability is of critical importance if it is to be used for long-term monitoring of the concentration changes of analytes through in vivo studies. We investigated the system stability using a saline solution containing 30 ng/mL of the respective analytes as test samples. The MD probe was inserted into the sample solution, and then on-line sampling and detection was conducted every 30 min for 5 h (10 consecutive measurements). We observed no decrease in the sensitivity or any deterioration of the precision ((15%, n ) 3) during the monitoring process, indicating that our on-line MDdesalter-ICPMS system was suited to prolonged analyses.

Application to in Vivo Studies. Measurement of Basal Values. We evaluated the applicability of our proposed on-line MDdesalter-ICPMS system for the simultaneous monitoring of multiple elements in the ECF of a living rat brain. As indicated in Table 2, the MDLs for the heavy elements were in the (sub)nanogram-per-milliliter regime, whereas those of Ca and Mg were in the range of 10-100 ng/mL. Figure 5 illustrates the concentration profiles obtained for the measurements of Cu, Zn, Mn, Mg, and Ca; we also attempted to simultaneously monitor the elements Co, Ni, Al, Mo, Pt, and Pb, but their concentrations in the ECF were too low to be determined by this system. As depicted in the Figure 5, each data point for Cu, Zn, and Mn represents an average concentration obtained over a 10 min interval (collected in a 10 µL sample loop), whereas those for Ca and Mg represent the results of continuous monitoring without prior collection in a sample loop. Figure 5 indicates that the concentration profiles for the tested elements exhibited similar features, except for that of Ca. A surge peak was observed ca. 10 min after insertion of the microprobe into the brain; subsequently, the concentration gradually decreased to an equilibrium level. The appearance of the peak differed for each analyte and even for the same analyte among different tested rats. This situation can be explained by considering the fact that insertion of the microprobe is an invasive process that inevitably ruptures nearby cells, causing the contents inside the cytosol to efflux out. After 2-3 h of sweeping the debris from the healthy cells and ECF, the system would then return to its basal state. It is interesting that Ca, in contrast to the other metal analyzed, did not exhibit a surging peak after insertion of the microprobe; rather, a steady increase in the concentration occurred until it reached an equilibrium level. The different behavior of Ca with respect to the other metals, Zn as a typical case, can be attributed to the fact that the intracellular concentration of Zn (150 µM) is much higher than its concentration in the CSF (0.15 µM);51,52 in contrast, the concentration of Ca in the cytosol is lower than that in the CSF.53 The basal concentration of each analyte was estimated 2-3 h after implantation of the MD probe to allow the environment to be restored to its equilibrium state. We obtained the following basal values obtained in this study (n ) 4): Zn, 5.22 ( 0.48 ng/ mL; Mn, 2.77 ( 0.43 ng/mL; Cu, 2.48 ( 0.38 ng/mL; Mg, 7.92 ( 1.79 µg/mL; Ca, 17.4 ( 2.29 µg/mL. Because there have been only a few studies concerning the measurement of trace elements in ECF, with different animal species used or different analytical techniques employed, it is difficult for us to directly compare our data with those in the literature. Nevertheless, our measured concentrations of Zn and Mn are in good agreement with literature values (Zn, 4-7 ng/mL;36,37 Mn, 0.83-1.50 ng/mL40,54). Our basal value of Cu (2.48 ( 0.38 ng/mL) differs significantly from the literature report (37 ng/mL), possibly because of the different animal species used or because the previous method had a high MDL (10 ng/mL).55 Only scattered data are available for Mg and Cu. The basal value of Mg measured by MD-GFAAS for male (51) Choi, D. W.; Koh, J. Y. Annu. Rev. Neurosci. 1998, 21, 347-375. (52) Takeda, A. Brain Res. Rev. 2000, 34, 137-148. (53) Gotoh, H.; Kajikawa, M.; Kato, H.; Suto, K. Brain Res. 1999, 828, 163168. (54) Takeda, A.; Sotogaku, N.; Oku, N. Brain Res. 2003, 965, 279-282. (55) Ma, H. M.; Wang, Z. H.; Zeng, Y.; Han, H. W.; Liu, G. Q. Chin. Chem. Lett. 1999, 10, 243-246.

Analytical Chemistry, Vol. 79, No. 23, December 1, 2007

8907

Figure 5. Time course of the concentration of trace metals in the ECF of a living rat following insertion of a microprobe into the brain: (a) Cu, Zn, and Mn with prior collection of microdialysate in a 10 µL loop; (b) Ca and Mg, real-time measurement without use of the sample loop. The error bars represent the standard deviations (n ) 3). The MD probe was inserted at time zero.

gerbils was reported to be 1.65 ng/mL, significantly lower than our value; the reported concentrations of ionized Mg (24.6-32.8 mg/L) and Ca (42.7-50 mg/L) in the CSF of humans are also significantly different from our values.56-58 Temporal Resolution of the System. The temporal resolution of microdialysis-based chemical monitoring is basically determined by the sample volume required for chemical separation (species separation by LC/CE or Na removal by a desalting system) and the sensitivity of the analytical method utilized.59,60 In our present on-line MD-membrane desalter-ICPMS system, MD is a continuous sampling process and separation of Na+ ions by the membrane desalter is based on a countercurrent exchange mechanism without any time delay; therefore, the limiting factor (56) Bogden, J. D.; Troiano, R. A.; Joselow, M. M. Clin. Chem. 1977, 23, 485489. (57) Burguera, J. L.; Burgurea, M.; Alarcon, O. M. J. Anal. At. Spectrom. 1986, 1, 79-83. (58) Basun, H.; Forssell, L. G.; Wetterberg, L.; Winblad, B. J. Neural Transm.: Parkinson’s Dis. Dementia Sect. 1991, 4, 231-258. (59) Zhou, S. Y.; Zuo, H.; Stobaugh, J. F.; Lunte, C. E.; Lunte, S. M. Anal. Chem. 1995, 67, 594-599. (60) Lada, M. W.; Kennedy, R. T. Anal. Chem. 1996, 68, 2790-2797.

8908 Analytical Chemistry, Vol. 79, No. 23, December 1, 2007

of the temporal resolution of the system is restricted by the detection sensitivity of the ICPMS. Because the concentrations of transition and other heavy metals in the ECF are usually very low [(sub)nanograms per milliliter], the microdialysate must be collected over a fixed time interval to provide a sufficient sample volume that meets the mass detection requirements of the ICPMS. As illustrated in Figure 1, we used a 10 µL sample loop to collect the microdialysate and diluent (EDTA solution), each at flow rate of 0.5 µL/min, over 10 min; thus, the temporal resolution for this system was 10 min. For Ca and Mg, which are present at much higher levels concentrations (micrograms per milliliter) in the ECF, we allowed the microdialysate to bypass the collection loop and flow directly to the desalter and subsequently to the ICPMS, resulting in near-real-time monitoring. The total delay in the response between the in vivo biochemical event and the output of the analytical data with this system was ca. 7.5 min. This delay mostly arose partly from the time required to move the microdialysate in the connection lines from the animal to the desalter and furthermore into the ICPMS, and partly from the traveling time that the microdialysate sample spent in the

Figure 6. Time course of (a) the Mn concentration and (b) the Pt signal in the ECF of a rat brain following intraperitoneal administration of drugs. Drug dosages: 100 mg MnCl2/kg and 4 mg cisplatin/kg body weight, respectively. The arrow (V) indicates the time at which the drug was injected. The MD probe was inserted at time zero.

desalter. Because of the very low internal volume (ca. 50 µL) of the CMD and the limited length of the associate connection tubing, the total delay time in the CMD would be less than 30 s. The total delay time of this system could possibly be minimized by shortening the length of its connection lines. Thus, our proposed on-line system is capable of performing near-real-time analyses of Ca and Mg, with temporal resolution as short as 10 min for the analyses of transition metals such as Cu, Mn, and Zn. Continuous and Dynamic Monitoring of Analytes. To demonstrate the performance of this system for dynamic monitoring, we performed experiments in which we dosed living rats with a high concentration of Mn and Pt exogenously through intraperitoneal (ip) administration. Mn, which is essential for normal brain functionality, moves across the blood-brain and blood-CSF barriers freely in the ionic form or when mediated by transferrin. In contrast, cisplatin, a well-known antineoplastic drug of hydrophilic low-mass Pt compounds, has difficulty penetrating the brain barrier. Figure 6 presents the concentration profiles of Mn and Pt, monitored for ca. 400 min after implantation of the MD probe to the animals. The administration of drugs was conducted at a time when the milieu near the MD probe had returned to its basal conditions. Figure 6a was obtained after dosing 100 mg MnCl2/ kg rat body weight (n ) 3). Following the administration of Mn, the average elapsed times for the initial rise and achieving the maximum value were ca. 24 and 72 min, respectively. The maximum concentration of Mn in the ECF approached 40 ng/ mL, significantly higher than its basal value (typically 1-2 ng/

mL). It is interesting to note that the concentration ceased to decrease in the time interval from 288 to 384 min but remained at a higher concentration (ca. 16 ng/mL) than the original basal value. The temporal profile for Pt after dosing with cisplatin (4 mg cisplatin/kg rat body weight) using the ip mode was also monitored using this on-line system. Figure 6b displays a typical profile among three animals tested; the y-axis in this figure represents the signal intensity. The initial rise in the Pt signal appeared ca. 26 min after administering cisplatin; it increased abruptly to a maximum level ca. 52 min later and then returned to its original basal value. This observation was unexpected because a literature report suggested that Pt could not be detected in the cerebral cortex-rich region (CCR) of cisplatin-injected mice without lipopolysaccharide (LPS) pretreatment.61 These contrasting observations might result from different animal species, dosages, modes of drug injection, or penetration mechanisms of the drug into the brain. Nevertheless, our in vivo and on-line continuous monitoring system should provide valuable assistance to further investigations regarding the use of this drug. CONCLUSION In comparison to the abundance of literature reports regarding the analysis of organic neurotransmitters such as aspartate and glutamate through in vivo microdialysis sampling with CE, LC, (61) Minami, T.; Okazaki, J.; Kawabata, A.; Kuroda, R.; Okazaki, Y. Toxicology 1998, 130, 107-113.

Analytical Chemistry, Vol. 79, No. 23, December 1, 2007

8909

or other techniques, so far there are only a few reports related to the in vivo measurement of trace elements. In this study we have successfully developed an on-line MD-membrane desalterICPMS system that is capable of performing in vivo measurements of multiple trace elements in the ECF of living animals. At a perfusion flow rate of 0.5 µL/min, the concentration recoveries of all of the elements tested reached 40-50% and remained unchanged even after long-term operation of the microdialysis system (up to 12 h). When using EDTA to complex with the metal ions and NH4Cl as a regenerant to exchange Na+ with NH4+ ions, the membrane desalter proved to be an effective system not only for quantitative removal of Na+ ions but also for preservation of metal analytes in the microdialysate sample. By taking advantage of the separation principles of the membrane suppressor, which is based on countercurrent exchange of NH4+ for Na+ supplied by the NH4Cl regenerant, the microdialysate obtained from microdialysis sampling could flow continuously through the desalting system without a time delay. This property can, therefore, provide a unique possibility for using this on-line system to achieve real-time monitoring. The temporal resolution of this system was determined by the MD flow rate, the separation of the salty matrix by the desalter, and the mass detection

8910

Analytical Chemistry, Vol. 79, No. 23, December 1, 2007

sensitivity of ICPMS. Under the present experimental conditions when using a fixed perfusion flow rate of 0.5 µL/min, the temporal resolution was limited by the mass sensitivity of the ICPMS. For elements such as Cu, Zn, and Mn that are present in the ECF at subnanogram-per-milliliter concentrations, the temporal resolution was 10 min because the microdialysate had to be collected initially using a sample loop to meet the sensitivity requirements of the ICPMS; for elements such as Ca and Mg that are present at microgram-per-milliliter concentrations, it was possible for our system to perform continuous monitoring that was close to real time. The combination of microdialysis and membrane desalter systems appears to be an especially powerful approach toward real-time monitoring of trace elements for in vivo studies. ACKNOWLEDGMENT One of the authors (M. H. Yang) gratefully acknowledges the financial support of the National Science Council of Taiwan for this study. Received for review May 15, 2007. Accepted September 10, 2007. AC070981Z