Anal. Chem. 2004, 76, 4123-4127
Studying Rat Brain Neurochemistry Using Nanoprobe NMR Spectroscopy: a Metabonomics Approach Purnima Khandelwal,† Chad E. Beyer,‡ Qian Lin,‡ Lee E. Schechter,‡ and Alvin C. Bach, II*,†
Chemical and Screening Sciences, Discovery Analytical Chemistry and Neuroscience Discovery Research, Wyeth Research, CN 8000, Princeton, New Jersey 08543-8000
In the present experiments, in vivo microdialysis techniques together with nanoprobe NMR spectroscopy were used to evaluate the neurochemical environment of the rat frontal cortex. Metabonomics techniques of data reduction and pattern recognition were used to examine whether collected neurochemicals were sensitive to tetrodotoxin (TTX), a neurotoxin that when infused into discrete brain regions can help distinguish between the neuronal versus glial origin of neurochemicals in cerebrospinal fluid microdialysate. 1H NMR spectra recorded on samples collected from the rat frontal cortex before and after an intracortical TTX infusion (10 µM for 60 min) were subjected to multivariate statistical analysis. Glutamate, isoleucine, valine, alanine, and r- and β-hydroxybutyrate were found to have decreased concentrations after the addition of TTX, suggesting that their release is likely from cortical neurons. In contrast, lactate, formate, acetate, glucose, creatinine, pyruvate, and other neurochemicals remained unchanged following local application of TTX. The present findings extend our previous work combining the analytical technology of small-volume nanoprobe NMR spectroscopy with in vivo microdialysis in freely moving animals and show that it is possible to apply metabonomics methodology to this important class of biofluid to monitor changes in neurochemical composition of the rat brain. Metabonomics can be defined as “multiparametric measurement” of changes in metabolism over time in response to a variety of stressors or interventions including but not limited to disease pathology, drugs, or toxins.1 One way in which the body responds to toxicity or disease is by altering biofluid composition. In this respect, various kinds of biofluids are used to carry out metabonomics studies. Urine2-5 and plasma6-10 are the most commonly used, since these are abundantly obtained in a minimally invasive manner and are appropriate for clinical trial monitoring, for disease diagnosis, and studies on animal models. On the other hand, * Corresponding author. E-mail:
[email protected]. † Chemical and Screening Sciences, Discovery Analytical Chemistry. ‡ Neuroscience Discovery Research. (1) Nicholson, J. K.; Connelly, J.; Lindon, J. C.; Holmes, E. Nat. Rev. Drug Discovery 2002, 1, 153-161. (2) Moolenaar, S. H.; Engelke, U. F.; Wevers, R. A. Ann. Clin. Biochem. 2003, 40, 16-24. 10.1021/ac049812u CCC: $27.50 Published on Web 05/29/2004
© 2004 American Chemical Society
cerebrospinal fluid (CSF), which is difficult to obtain (i.e., it requires lumbar puncture) and not generally available in abundant quantities, is the most useful biofluid to study neuropathology and neuropsychiatric disorders. The difficulty of studying CSF is further supported by the fact that the last CSF application to metabonomics was 10 years ago.11 It has not proved to be a popular biofluid in a research environment because study of the time course of metabolic changes is not possible with an invasive method of lumbar puncture. To address some of the issues regarding studies with CSF, brain microdialysis,12,13 a very popular (more than 800 published reports each year) technique that samples the extracellular concentration of neurochemicals, can be employed. In this method, dialysis probes containing a semipermeable membrane are implanted in discrete brain regions of freely moving animals. The most commonly used probe is a concentric-style probe consisting of two tubes, one inside the other, as shown in a schematic in Figure 1. A perfusate containing minimal salts is passed through the inner tube from where it flows out into the outer tube that is in contact with the extracellular environment. This allows for an exchange of metabolites/neurochemicals through the semipermeable membrane for subsequent collection and analysis. This method makes the neurochemical (3) Lindon, J. C.; Nicholson, J. K.; Holmes, E.; Antti, H.; Bollard, M. E.; Keun, H.; Beckonert, O.; Ebbels, T. M.; Reily, M. D.; Robertson, D.; Stevens, G. J.; Luke, P.; Breau, A. P.; Cantor, G. H.; Bible, R. H.; Niederhauser, U.; Senn, H.; Schlotterbeck, G.; Sidelmann, U. G.; Laursen, S. M.; Tymiak, A.; Car, B. D.; Lehman-McKeeman, L.; Colet, J. M.; Loukaci, A.; Thomas, C. Toxicol. Appl. Pharmacol. 2003, 187, 137-146. (4) Shockcor, J. P.; Holmes, E. Curr. Top. Med. Chem. 2002, 2, 35-51. (5) Keun, H. C.; Beckonert, O.; Griffin, J. L.; Richter, C.; Moskau, D.; Lindon, J. C.; Nicholson, J. K. Anal. Chem. 2002, 74, 4588-4593. (6) Coen, M.; Lenz, E. M.; Nicholson, J. K.; Wilson, I. D.; Pognan, F.; Lindon, J. C. Chem. Res. Toxicol. 2003, 16, 295-303. (7) Griffin, J. L.; Walker, L. A.; Shore, R. F.; Nicholson, J. K. Chem. Res. Toxicol. 2001, 14, 1428-1434. (8) Waters, N. J.; Holmes, E.; Williams, A.; Waterfield, C. J.; Farrant, R. D.; Nicholson, J. K. Chem. Res. Toxicol. 2001, 14, 1401-1412. (9) Nicholls, A. W.; Holmes, E.; Lindon, J. C.; Shockcor, J. P.; Farrant, R. D.; Haselden, J. N.; Damment, S. J.; Waterfield, C. J.; Nicholson, J. K. Chem. Res. Toxicol. 2001, 14, 975-987. (10) Griffin, J. L.; Walker, L. A.; Garrod, S.; Holmes, E.; Shore, R. F.; Nicholson, J. K. Comp. Biochem. Physiol. B 2000, 127, 357-367. (11) Ghauri, F. Y.; Nicholson, J. K.; Sweatman, B. C.; Wood, J.; Beddell, C. R.; Lindon, J. C.; Cairns, N. J. NMR Biomed. 1993, 6, 163-167. (12) Ungerstedt, U. In Techniques in the behavioral and neural sciences: microdialysis in the neurosciences; Robinson, T. E., Justice, J. B., Jr., Eds.; Elsevier: Amsterdam, 1991; Vol. 7, pp 3-18. (13) Jamal, M.; Ameno, K.; Kumihashi, M.; Ameno, S.; Kubota, T.; Wang, W.; Ijiri, I. J. Chromatogr., B 2003, 798, 155-158.
Analytical Chemistry, Vol. 76, No. 14, July 15, 2004 4123
Figure 2. Chemical structure of the neurotoxin tetrodotoxin (TTX) in its zwitterionic form.
Figure 1. Schematic showing a concentric-style microdialysis probe containing a semipermeable membrane at one end. It is surgically implanted in discrete brain regions (frontal cortex in the present case) of freely moving rats.
environment more amenable to analytical studies, since it is possible to measure basal, stimulated, and drug-induced changes in extracellular neurochemical concentrations in almost real time. Many analytical methods have been coupled with the microdialysis technique to enable monitoring of various neurochemicals, for example, high performance liquid chromatography with electrochemical detection,14 mass spectrometry,15 gas chromatography,16 capillary electrophoresis,17 ion-exchange chromatography18 and immunosensor analysis.19 Additionally, nanoprobe NMR spectroscopy has recently been shown to be very useful in studying neurochemicals collected from microdialysis experiments.20 This latter combination offers many advantages over other analytical methods. For example, nanoprobe NMR spectroscopy allows simultaneous analysis of all chemical components present above detectable levels, can detect unanticipated or druginduced changes in biofluids, and is nondestructive and rapid while requiring small sample volumes. Sample preparation for NMR spectroscopy is quite straightforward and does not need any pretreatment or derivatization, making it possible to study chemical components in their natural state and concentrations. Statistical methods, such as principal components analysis (PCA),21 in conjunction with NMR spectroscopy help in easy visualization of similarities and differences in multiple samples by identifying inherent patterns in the NMR spectra. The method entails reduction of data by division of the NMR spectrum into equal chemical shifts regions, followed by signal integration within those ranges to generate a set of variables. Subsequently, PCA is used to summarize this data table with multiple variables by creating a few new variables containing most of the information. The new (14) Beyer, C. E.; Boikess, S.; Luo, B.; Dawson, L. A. J. Psychopharmacol. 2002, 16, 297-304. (15) Hows, M. E.; Organ, A. J.; Murray, S.; Dawson, L. A.; Foxton, R.; Heidbreder, C.; Hughes, Z. A.; Lacroix, L.; Shah, A. J. J. Neurosci. Methods 2002, 121, 33-39. (16) Kondrat, R. W.; Kanamori, K.; Ross, B. D. J. Neurosci. Methods 2002, 120, 179-192. (17) Li, Y. M.; Qu, Y.; Vandenbussche, E.; Arckens, L.; Vandesande, F. J. Neurosci. Methods 2001, 105, 211-215. (18) Heidbreder, C. A.; Lacroix, L.; Atkins, A. R.; Organ, A. J.; Murray, S.; West, A.; Shah, A. J. J. Neurosci. Methods 2001, 112, 135-144. (19) Cook, C. J. J. Neurosci. Methods 2001, 110, 95-101. (20) Khandelwal, P.; Beyer, C. E.; Lin, Q.; McGonigle, P.; Schechter, L. E.; Bach, A. C., 2nd. J. Neurosci. Methods 2004, 133, 181-189. (21) Lindon, J. C.; Holmes, E.; Nicholson, J. K. Prog. Nuclear Magn. Reson. Spectrosc. 2001, 39, 1-40.
4124 Analytical Chemistry, Vol. 76, No. 14, July 15, 2004
variables are computed as a linear combination of all the variables and are called scores (denoted by t).22 The weightings with which the variables are multiplied to form the scores are called loadings that signify the importance of these variables. In the present report, we apply the techniques of nanoprobe NMR spectroscopy, brain microdialysis, and multivariate metabonomics to study the neurochemical constituents of the rat frontal cortex before and after local infusion of the well-studied neurotoxin, TTX23-25 (Figure 2). This neurotoxin, by blocking the diffusion of sodium ions through ion channels located on presynaptic neurons, enables us to determine whether a particular neurochemical is released from neurons. Thus, a local TTX infusion prevents the propagation of action potentials in neurons, thereby blocking release of neurochemicals in the extracellular space that are collected by microdialysis studies. EXPERIMENTAL SECTION Animal Experiments. Animal studies were performed in accordance to Wyeth’s Internal Animal Care and Use Committee and the Guide for the Care and Use of Laboratory Animals as promulgated by the National Institutes of Health (Publication 8523, 1985). Eight male Sprague-Dawley rats (Charles River, Wilmington, MA) weighing 280-350 g at the time of surgery were used for all experiments. Prior to surgery, animals had free access to food and water and were group-housed in an AAALC-accredited facility that was maintained on a 12-h light/dark cycle (lights on at 0600 h). All experiments were done during the light phase. Surgical Procedures. Following induction of anesthesia with gaseous administration of 3% halothane (Fluothane; Zeneca, Cheshire, U.K.), animals were secured in a stereotaxic frame with ear and incisor bars (David Kopf, Tujunga, CA). A microdialysis guide cannula (CMA/12, CMA/Microdialysis, Stockholm, Sweden) was directed toward the frontal cortex in the brain. The coordinates for this surgery were: +3.2 mm anterior to bregma, -3.5 mm lateral from the midline and -1.3 mm ventral to dura with a flat skull.26 Guide cannulae were secured to the skull using dental acrylic (Plastics One, Roanoke, VA) and two stainless steel screws. Following surgery, the animals were individually housed in Plexiglass cages (45 cm sq) and were provided food and water ad libitum. Animals were allowed ∼24 h of post-operative recovery time. (22) Eriksson, L.; Johansson, E.; Kettaneh-Wold, N.; Wold, S. Multi- and Megavariate Data Analysis: Principles and Applications; Umetrics Academy: Umea, Sweden, 2001. (23) Westerink, B. H.; De Vries, J. B. J. Neurochem. 1988, 51, 683-687. (24) Fuchs, H.; Hauber, W. Exp. Brain Res. 2004, 154, 66-75. (25) Timmerman, W.; Westerink, B. H. Synapse 1997, 27, 242-261. (26) Paxinos, G.; Watson, C. The rat brain in stereotaxic coordinates; Academic Press: New York, 1986.
In vivo Microdialysis. Microdialysis probes ( o.d., 0.5 mm) were purchased from CMA/Microdialysis (Stockholm, Sweden) and were equilibrated according to the manufacturer’s specifications. The active length of the microdialysis probe was 2 mm. Initially, microdialysis probes were perfused with artificial CSF (aCSF; 125 mM NaCl, 3 mM KCl, 0.75 mM MgSO4, and 1.2 mM CaCl2, pH 7.4) in a glass beaker for at least 18 h prior to experimentation. The microdialysis probe was then implanted via the guide cannula into the frontal cortex and perfused with aCSF at a flow rate of 1 µL/min. After the probe was inserted into the animal, a 3-h stabilization period was allowed before dialysate samples were collected in the unrestrained rat. After this stabilization period, five baseline samples were collected in 20-min intervals to establish a steady baseline, after which TTX (10 µM) was infused for 60 min through the probe into the frontal cortex (CMA/110 liquid switch system). Samples were collected for 2 h in 20-µL fractions after TTX infusion was complete. Sodium azide (0.1%) was added to the CSF microdialysate tubes immediately after collection and placed on dry ice. At the end of the experiment, animals were euthanized, and the probe placement was verified histologically. Animals with incorrect probe placement were excluded from the study. NMR Experiments. Solvent was completely removed by speed-vacuuming on a Savant SC 110A (Savant Instruments, Inc. Holbrook, NY). A 20-µL portion of solvent containing 5 µM d4TSP (2,2,3,3-d4 sodium 3-(trimethylsilyl)propionate) in D2O was added to prepare the NMR sample, which was transferred to a Varian (Palo Alto, CA) nanoprobe cell. All chemicals used were analytical grade and were purchased from Sigma-Aldrich chemicals (Milwaukee, WI). d4-TSP and D2O were purchased from Isotec, a member of Sigma-Aldrich. The sample was spun at 2 kHz at 25 °C. All the NMR experiments were recorded on a 600MHz Varian Unity INOVA spectrometer (Palo Alto, CA), using a GHX nanoprobe fitted with z-gradients. 1H 1D experiments were recorded on five 20-µL samples collected before and five collected after TTX infusion from each rat. Experimental parameters were as follows: 10 ppm spectral width; 3.5-s recycle delay, including 1.5-s presaturation time (low power, suitable for saturation of residual water, was used); 2-k scans. Each 1D experiment took ∼2 h. The spectra were processed using Varian’s VNMR program. An exponential line broadening of 1.0 Hz was used. Multivariate Statistical Analysis. Integrated intensities of 0.02 ppm regions of 1H 1D spectra were used in the SIMCA-P 9.0 (Umetrics, Kinnelon, NJ) multivariate data analysis software to carry out PCA on microdialysate samples from the frontal cortex collected before and after TTX infusion. A total of 24 spectra before TTX and 24 after TTX were used. The region where the water resonance occurs in the spectrum, 4.75-5.04 ppm, was not included in the calculations to avoid any spurious effects caused by variability in the suppression of the residual water resonance. Moreover, spinning sidebands as well as 0.2 ppm sections of spectra containing impurity peaks (ethanol, 1.11, 3.61 ppm; acetone, 2.24 ppm) were also excluded from the analysis. Each data point was normalized to the sum of its row and all variables were mean-centered prior to PCA. Scores plots of the principal components were constructed to visualize separation of the spectra on the basis of the time of collection, that is, before or after TTX infusion. From the values of the PC loadings, which indicate the
Figure 3. NMR spectra from 0.6 to 3.0 ppm of frontal cortex microdialysate before and after 40 min of TTX infusion. Neurochemicals undergoing change in concentration are labeled in the “before TTX” spectrum, whereas peaks that remain unchanged in intensity are labeled in the “after TTX” NMR spectrum. Artifacts arising from spinning sidebands are marked with an asterisk.
importance of each variable to the clustering, key NMR spectral regions that change upon TTX addition were identified. RESULTS AND DISCUSSION In a recent study,20 our research group described methods used to assign NMR spectra and identified more than 20 neurochemicals from five regions of the rat brain. In the present report, we extended our previous efforts to measure concentration changes in neurochemicals after infusion with TTX. To carry out these experiments, five 20-µL (at a rate of 1 µL/min) microdialysate samples were collected before and five after TTX infusion from each rat. 1H 1D experiments were carried out on these collected samples. Figure 3 shows NMR spectra from 0.6 to 3.0 ppm of frontal cortex microdialysate before the infusion of TTX and a second microdialysate sample after 40 min of TTX infusion. Signals arising from Glu β-protons (2.00, 2.36 ppm), Ala methyl (1.42 ppm), Ile (0.94 and 1.01 ppm), and R-hydroxybutyric acid (0.90 ppm) are significantly reduced in intensity after TTX infusion, whereas those arising from lactate, pyruvate, and acetate have the same intensity in both the before TTX and after TTX spectra. When other regions of NMR spectra were inspected, for example from 3.0 to 4.0 ppm (Figure 4), where most of the peaks arising from R- and β-glucose are present, very few changes were seen. Since this is a very crowded region of the spectrum, some changes might not have been obvious by visual inspection. To study the obvious and nonobvious differences in the NMR spectra from before and after TTX infusion samples, the multivariate analysis technique called principal components analysis was utilized. Figure 5a shows the scores plot for the samples collected before (red) and after TTX infusion (blue). Two principal components (t1 and t2) that explain 67% of the variation in the data Analytical Chemistry, Vol. 76, No. 14, July 15, 2004
4125
Figure 4. 3.0 to 4.0 ppm region of the NMR spectra from frontal cortex microdialysate before and after 40 min of TTX infusion. Peaks marked with an asterisk are artifacts caused by spinning of the sample in the nanoprobe.
were used to obtain this plot. The plot depicts that the two kinds of microdialysate samples are present in opposite corners of the plot, and a separation between the two is very clear. The points on the scores plot are labeled with the respective rat number (a to h) and the time (in multiples of 20 min) at which they were collected. b12 is seen as an outlier because the NMR spectrum of this sample showed some spurious peaks. There is some overlap of before and after TTX points, especially from rats A and F. The NMR spectra from these rats did not show the same changes observed in the spectra obtained from the other rats in the study. The reason for this anomaly is unclear but could be due to variability between animals or the result of nonideal TTX infusion. To find out which regions of NMR spectra cause the separation of the before and after TTX points in the scores plot, we constructed a loadings plot by using the weightings. Figure 5b shows the loadings plot for the data. Neurochemical peaks with significantly altered levels among the before and after samples appear at the same region in the scores and the loadings plot, and the most important variables are found in the peripheral parts of the loadings plot. Using these criteria, we drew an ellipse containing the variables (“bucket” integrals) responsible for the separation seen in the scores plot. Most of the variables lying inside the ellipse come from the same region as was seen in Figure 3, for example, 2.00, 2.36, 0.91, 1.01, and 1.07 ppm. These variables can be traced to individual peaks in the NMR spectrum arising from specific neurochemicals, listed in Table 1. It is observed that changes in concentrations of only a few neurochemical species lead to the differences observed in the scores plot. And since concentration changes occur after TTX infusion for only those compounds that are released from neurons, the following can be identified as having a neuronal origin: R-hydroxyisovalerate, R-hydroxybutyrate, isoleucine, valine, alanine, 4126 Analytical Chemistry, Vol. 76, No. 14, July 15, 2004
Figure 5. (a) Scores plot of the first two principal components obtained using SIMCA (see text). Integrated intensities from the whole spectrum (0.2-10.0 ppm) were used to obtain the plot. Thirty-two microdialysate samples from six rats, collected before (red) and after (blue) TTX infusion were included in the PCA analysis (see text for details). (b) Loadings plot for the first two principal components. The most important variables are found in the peripheral parts of the loadings plot, present here inside the green ellipse. Variables shown in red in the central cluster arise from neurochemical peaks that remain unchanged after TTX infusion as described in the text. Table 1. Identification in the Loadings Plot of Variables That Are Responsible for Differences Observed between the before and after TTX Infusion Spectra variable in loadings plot
proton assignment (multiplicitya)
neurochemical
0.83 0.91 0.95 0.99 1.01 1.07 1.21 1.27 1.41 1.98 2.00 2.36
CH3 (d) CH3 (t) δ-CH3 (t) CH3 (d) β-CH3 (d) CH3 (d) CH3 (d) γ-CH2 (m) CH3 (d) β-CH (m) β-CH2 (m) γ-CH2 (m)
R-hydroxyisovalerate R-hydroxybutyrate isoleucine valine isoleucine valine β-hydroxybutyrate isoleucine alanine isoleucine glutamate glutamate
a
d ) doublet, t ) triplet, m ) multiplet.
glutamate, and β-hydroxybutyrate. This agrees with earlier studies on this region of the brain that also identify these amino acids as arising from the neurons, although using other analytical techniques for measurement.27-29 On the other hand, other neurochemicals, such as glucose, myoinositol, creatinine, formate, (27) Frantz, K.; Harte, M.; Ungerstedt, U.; O’Connor, W. T. J. Neurosci. Methods 2002, 119, 109-119.
acetate, and pyruvate. are essentially unchanged upon TTX infusion, meaning they originate from the glial cells. This is seen in the loadings plot in Figure 5b where integrals for regions containing peaks for glucose (5.25 ppm), dimethylamine (2.76 ppm), formate (8.46 ppm), creatinine (3.05 ppm), and pyruvate (2.38 ppm) (shown in red) are present in the central cluster, which signifies the unchanged/insignificant part between the before and after TTX spectra. NMR spectroscopy has been especially useful in this case, since we have been able to identify a greater number of neurochemicals in the same set of experiments. Specialized methods set up to measure amino acids are often not sensitive to other classes of molecules.15,27,29-31 NMR spectroscopy has the advantage of being a “universal detector” of protons, compensating for its poor sensitivity compared to HPLC. CONCLUSIONS Using the technique of nanoprobe NMR spectroscopy, we were able to measure changes in the composition of neurochemicals in the rat brain after infusion with a neurotoxin, TTX. Until now, other techniques, such as HPLC, have been used to monitor neurochemical levels in collected microdialysate; however, these techniques are generally able to focus on a few neurochemicals at a given time, and the samples often need modifications (i.e., derivatization) and cannot usually measure unanticipated changes. (28) Dawson, L. A.; Nguyen, H. Q.; Li, P. Neuropsychopharmacology 2001, 25, 662-668. (29) Timmerman, W.; Cisci, G.; Nap, A.; de Vries, J. B.; Westerink, B. H. Brain Res. 1999, 833, 150-160. (30) Bianchi, L.; Ballini, C.; Colivicchi, M. A.; Della Corte, L.; Giovannini, M. G.; Pepeu, G. Neurochem. Res. 2003, 28, 565-573. (31) Yoshitake, T.; Iizuka, R.; Kehr, J.; Nohta, H.; Ishida, J.; Yamaguchi, M. J. Neurosci. Methods 2001, 109, 91-96.
With the present technique, it is possible to monitor global changes in all the neurochemicals above limits of detection (low micromolar) simultaneously using NMR spectroscopy as a universal detector. This method can be advantageous in many situations when a new model of a disease is being developed or new kinds of biomarkers for existing models need to be identified through metabonomics. It is also possible to apply a combination of microdialysis and nanoprobe NMR spectroscopy to other sites that can be sampled simultaneously with microdialysis, for example, skin,32,33 ocular,34,35 heart,36 tumor,37,38 and synovial fluid.39,40 ACKNOWLEDGMENT The authors are grateful to Juan Mercado and Sue Asbury for their expert surgical assistance. The authors also thank Drs. Oliver McConnell and Sharon Rosenzweig-Lipson for their support of this work. Received for review February 3, 2004. Accepted April 26, 2004. AC049812U (32) Bielecka-Grzela, S.; Klimowicz, A. J. Clin. Pharm. Ther. 2003, 28, 465469. (33) McDonald, S.; Lunte, C. Pharm. Res. 2003, 20, 1827-1834. (34) Macha, S.; Mitra, A. K. J. Ocul. Pharmacol. Ther. 2001, 17, 485-498. (35) Rittenhouse, K. D.; Pollack, G. M. Adv. Drug Delivery Rev. 2000, 45, 229241. (36) Davani, S.; Chocron, S.; Muret, P.; Mersin, N.; Etievent, J. P.; Kantelip, J. P. Pathol. Biol. (Paris) 2003, 51, 39-43. (37) Brunner, M.; Muller, M. Eur. J. Clin. Pharmacol. 2002, 58, 227-234. (38) Johansen, M. J.; Thapar, N.; Newman, R. A.; Madden, T. J. Exp. Ther. Oncol. 2002, 2, 163-173. (39) Lawand, N. B.; McNearney, T.; Westlund, K. N. Pain 2000, 86, 69-74. (40) Qian, M.; West, W.; Wu, J. T.; Lu, B.; Christ, D. D. Pharm. Res. 2003, 20, 605-610.
Analytical Chemistry, Vol. 76, No. 14, July 15, 2004
4127
