Imaging of l-Glutamate Fluxes in Mouse Brain Slices Based on an

Jul 2, 2003 - A time-resolved imaging method for visualizing l-glutamate release in mammalian brain slices is proposed by using an enzyme membrane com...
0 downloads 6 Views 693KB Size
Anal. Chem. 2003, 75, 3775-3783

Imaging of L-Glutamate Fluxes in Mouse Brain Slices Based on an Enzyme-Based Membrane Combined with a Difference-Image Analysis Ayumi Hirano,*,† Nobukatsu Moridera,† Mai Akashi,† Minoru Saito,† and Masao Sugawara‡

Department of Physics and Applied Physics and Department of Chemistry, College of Humanities and Sciences, Nihon University, Sakurajousui, Setagaya, Tokyo 156-8550, Japan

A time-resolved imaging method for visualizing L-glutamate release in mammalian brain slices is proposed by using an enzyme membrane combined with a difference-image analysis. The enzyme membrane is composed of Lglutamate oxidase and horseradish peroxidase incorporated into a bovine serum albumin matrix. L-Glutamate triggers an enzyme-coupling reaction to convert a redox substrate (DA-64) to Bindschedler’s Green, which gives a green color signal. The difference-image analysis is based on calculating slopes of a signal versus time (t) plot in the time range from (t - 40 s) to (t + 40 s) for visualizing L-glutamate release in terms of its flux (in mol min-1 cm-2). The method was applied to a time-resolved imaging of hippocampal distribution of ischemia-induced L-glutamate release in mouse brain slices. The image of L-glutamate distribution showed that the level and time courses of L-glutamate fluxes were neuronal regiondependent. The maximum flux of L-glutamate at CA1 was observed at 7.7 min after ischemia. The flux at 7.7 min increased in the order of CA1 ≈ CA3 > DG. The time course of the L-glutamate flux in the CA1 region was biphasic and that in the DG region was modestly biphasic. In the CA3 region, such biphasic release of L-glutamate was not seen. The ischemia-induced L-glutamate flux was accelerated when Mg2+ was omitted from an extracellular solution. The present enzyme membrane-based approach provides a useful method for visualizing distribution of L-glutamate release in the brain slices during ischemia. Imaging of the spatial and time-resolved distribution of biologically active substances in cells and tissues has attracted much attention in view of obtaining direct information about intracellular and intercellular processes of biological signaling cascades.1 Many studies have been described on the imaging of dynamic changes in the spatial distribution of intracellular messengers, such as Ca2+,2-4 cyclic AMP,5,6 cyclic GMP,7 NO,8,9 and protein kinase C.10 Intracellular signaling events, such as protein phosphorylation11 * To whom all correspondence should be addressed. Tel: [81] (03) 33291151. Fax: [81] (03) 5317-9432. E-mail: [email protected]. † Department of Physics and Applied Physics. ‡ Department of Chemistry. (1) Tsien, R. Y. Annu. Rev. Neurosci. 1989, 12, 227-253. (2) Regehr, W. G.; Connor, J. A.; Tank, D. W. Nature 1989, 341, 533-536. (3) Regehr, W. G.; Tank, D. W. Nature 1990, 345, 807-810. 10.1021/ac030088+ CCC: $25.00 Published on Web 07/02/2003

© 2003 American Chemical Society

and conformational changes of intracellular proteins12 and proteinprotein interaction,13-15 have also been visualized in living cells. The static distribution of membrane proteins, such as the N-methyl-D-aspartate (NMDA) receptor and its subunits in brain slices, has been imaged by autoradiography,16,17 immunohistochemistry,18,19 and in situ hybridization.20-22 However, a limited number of studies have been reported on visualizing the distribution of extracellular small molecules, for example, neurotransmitters.23-29 (4) Miyawaki, A.; Llopis, J.; Heim, R.; MaCaffery, J. M.; Adams, J. A.; Ikura, M.; Tsien, R. Y. Nature 1997, 388, 882-887. (5) Adams, S. R.; Harootunian, A. T.; Buechler, Y. J.; Taylor, S. S.; Tsien, R. Y. Nature 1991, 349, 694-697. (6) Hempel, C. M.; Vincent, P.; Adams, S. R.; Tsien, R. Y.; Selverston, A. I. Nature 1996, 384, 166-169. (7) Sato, M.; Hida, N.; Ozawa, T.; Umezawa, Y. Anal. Chem. 2000, 72, 59185924. (8) Kojima, H.; Nakatsubo, N.; Kikuchi, K.; Kawahara, S.; Kirino, Y.; Nagoshi, H.; Hirata, Y.; Nagano, T. Anal. Chem. 1998, 70, 2446-2453. (9) Kojima, H.; Hirata, M.; Kudo, Y.; Kikuchi, K.; Nagano, T. J. Neurochem. 2001, 76, 1404-1410. (10) Chen, C.-S.; Poenie, M. J. Biol. Chem. 1993, 268, 15812-15822. (11) Sato, M.; Ozawa, T.; Inukai, K.; Asano, T.; Umezawa, Y. Nat. Biotechnol. 2002, 20, 287-294. (12) Nakanishi, J.; Nakajima, T.; Sato, M.; Ozawa, T.; Tohda, K.; Umezawa, Y. Anal. Chem. 2001, 73, 2920-2928. (13) Ozawa, T.; Nogami, S.; Sato, M.; Ohya, Y,; Umezawa, Y. Anal. Chem. 2000, 72, 5151-5157. (14) Ozawa, T.; Kaihara, A.; Sato, M.; Tachihara, K.; Umezawa, Y. Anal. Chem. 2001, 73, 2516-2521. (15) Ozawa, T.; Takeushi, T. M.; Kaihara, A.; Sato, M.; Umezawa, Y. Anal. Chem. 2001, 73, 5866-5874. (16) Monaghan, D. T.; Cotman, C. W. J. Neurosci. 1985, 5, 2909-2919. (17) Laurie, D. J.; Seeburg, P. H. Eur. J. Pharmacol., 1994, 268, 335-345. (18) Wenzel, A.; Scheurer, L.; Ku ¨ nzi, R.; Fritschy, J. M.; Mohler, H.; Benke, D. NeuroReport 1995, 7, 45-48. (19) Watanabe, M.; Fukaya, M.; Sakimura, K.; Manabe, T.; Mishina, M.; Inoue. Y. Eur. J. Neurosci. 1998, 10, 478-487. (20) Watanabe, M.; Inoue, Y.; Sakimura, K.; Mishina, M. J. Comp. Neurol. 1993, 338, 377-390. (21) Kutsuwada, T.; Kashiwabuchi, N.; Mori, H.; Sakimura, K.; Kushiya, E.; Araki, K.; Meguro, H.; Masaki, H.; Kumanishi, T.; Arakawa. M.; Mishina, M. Nature 1992, 358, 36-41. (22) Watanabe, M.; Inoue, Y.; Sakimura, K.; Mishina, M. NeuroReport 1992, 3, 1138-1140. (23) Tan, W.; Haydon, P. G.; Yeung, E. S. Appl. Spectrosc. 1997, 51, 11391143. (24) Parpura, V.; Tong, W.; Yeung, E. S.; Haydon, P. G. J. Neurosci. Methods 1998, 82, 151-158. (25) Lillard, S. J.; Yeung, E. S. J. Neurosci. Methods 1997, 75, 103-109. (26) Tong, W.; Yeung, E. S. Appl. Spectrosc. 1998, 52, 407-413. (27) Qian, W.-J.; Aspinwall, C. A.; Battiste, M. A.; Kennedy, R. T. Anal. Chem. 2000, 72, 711-717.

Analytical Chemistry, Vol. 75, No. 15, August 1, 2003 3775

L-Glutamate is a principal neurotransmitter in brain and plays a key role in synaptic plasticity, memory formation, and neurotoxicity.30,31 Visualization of regional distribution of L-glutamate release in brains or brain slices is important for understanding physiological and pathological actions of L-glutamate, because neurons in different regions, for example, cornu ammonis 1 (CA1), CA3, and dentate gyrus (DG), in the hippocampus exhibit different responses to the same stimulation.32,33 Several approaches have been reported for imaging of L-glutamate released from brain (hippocampal) slices. An approach based on an enzyme reaction between L-glutamate and glutamate dehydrogenase (GDH) in the presence of NAD to produce fluorescent NADH34-38 has been used for fluorometric imaging of L-glutamate released from gerbil hippocampal slices.34 A clonal cell monolayer of fura 2-loaded Chinese hamster ovary (CHO) cell lines expressing NMDA receptor has been utilized as a sensory device for visualizing L-glutamate released from rat hippocampal slices.39 An electrochemical imaging method using an electrode array modified with L-glutamate oxidase (GluOx) and horseradish peroxidase (HRP) has also been reported for monitoring of L-glutamate release at multiple neuronal positions in rat hippocampal slices.40,41 The spatial resolution of the above approaches is, however, not sufficient for detecting a regional difference in L-glutamate release, due to high level of basal fluorescence from endogeneous NADH, diffusional dispersion of L-glutamate during superfusing extracellular media, or both. In the electrochemical approach, the spatial resolution is limited by overlapping of diffusion layers at respective electrodes.42 For enhancement of the spatial resolution of L-glutamate released in brain slices, we have proposed a nonsuperfusion approach by using a GluOx-HRP membrane and a nonfluorescent dye DA-64 (substrate).43 To supply the enzymes into the extracellular fluid of brain slices without superfusion, an enzyme membrane was prepared by incorporating GluOx and HRP into a bovine serum albumin (BSA) matrix on a poly(L-lysine)-coated cover slip by the glutaraldehyde method. When a brain slice is loaded with DA-64 and placed on the membrane, the enzymes diffuse from the membrane into the extracellular fluid of the brain slice, where L-glutamate released from neurons triggers an enzyme reaction to convert DA-64 to Bindschedler’s Green (BG). The

(28) Gee, K. R.; Zhou, Z.-L.; Qian, W.-J.; Kennedy, R. J. Am. Chem. Soc. 2002, 124, 776-778. (29) Wang, Z.; Haydon, P. G.; Yeung, E. S. Anal. Chem. 2000, 72, 2001-2007. (30) Bliss, T. V. P.; Collingridge, G. L. Nature 1993, 361, 31-39. (31) Nishizawa, Y. Life Sci. 2001, 69, 369-381. (32) Schmidt-Kastner, R.; Freund, T. F. Neuroscience 1991, 40, 599-636. (33) Manabe, T. Rev. Neurosci. 1998, 8, 179-193. (34) Mitani, A.; Kadoya, F.; Nakamura, Y.; Kataoka, K. Neurosci. Lett. 1991, 122, 167-170. (35) Ayoub, G. S.; Dorst, K. Vision Res. 1998, 38, 2909-2912. (36) Ayoub, G. S.; Grutsis, S.; Simko, H. J. Neurosci. Methods 1998, 81, 113119. (37) Tohda, C.; Kuraishi, Y. Neurosci. Res. 1996, 24, 183-187. (38) Innocenti, B.; Parpura, V.; Haydon, P. G. J. Neurosci. 2000, 20, 1800-1808. (39) Uchino, S.; Nakamura, T.; Nakamura, K.; Nakajima-Iijima, S.; Mishina, M.; Kohsaka, S.; Kudo, Y. Eur. J. Neurosci. 2001, 13, 670-678. (40) Kasai, N.; Jimbo, Y.; Niwa, O.; Matsue, T.; Torimitsu, K. Neurosci. Lett. 2001, 304, 112-116. (41) Kasai, N.; Jimbo, Y.; Torimitsu, K. Anal. Sci. 2002, 18, 1325-1327. (42) Hayashi, K.; Horiuchi, T.; Kurita, R.; Torimitsu, K.; Niwa, O. Biosens. Bioelectron. 2000, 15, 523-529. (43) Hirano, A.; Asakawa, M.; Kido, N.; Sugawara, M. Anal. Sci. 2000, 16, 2529.

3776 Analytical Chemistry, Vol. 75, No. 15, August 1, 2003

Figure 1. Schematic representation of a time-resolved imaging method for L-glutamate fluxes.

regional distribution of KCl-induced L-glutamate release in the hippocampus of mouse brain slices was visualized by using the GluOx-HRP membranes. However, the enzyme reaction was irreversible and consumed L-glutamate, forming BG irreversibly. The intensity of green signal due to BG does not disappear even when L-glutamate release was stopped. Therefore, the visual image reflected the time-integrated amount of L-glutamate release. To monitor time courses of chemical reactions, a time-resolved difference spectrum is an excellent way to extract a transient process simply by subtracting spectra obtained at a given time interval.44 Ayoub et al. introduced a difference-image approach for obtaining a time-resolved fluorometric image of L-glutamate release based on GDH and NAD.35,36 For each image evaluated, the previous image was digitally subtracted. On the basis of this approach, they visualized the localization of recent L-glutamate release in retinal slices from zebra fish under basal and elevated potassium conditions. In the present study, we combine a difference-image approach with the GluOx-HRP membrane method for time-resolved imaging of hippocampal distribution of L-glutamate release in mouse brain slices. The principle is shown in Figure 1. A color image is taken by a digital camera at a given time interval, and the obtained image is first divided into small regions, named as regions of interest (ROIs). Next, the mean intensity of the digital color signal, in the present case that (R(t)) of red component, is calculated for each ROI and plotted against time t. The slope of the R(t) versus t plot has a dimension of mol min-1 cm-2 and is a measure of a L-glutamate flux at each measuring point (ROI). The calculation (44) Chen, E.; Goldbeck, R. A.; Kliger, D. S. Annu. Rev. Biophys. Biomol. Struct. 1997, 26, 327-355.

of slopes for all ROIs yields a two-dimensional visual image for L-glutamate fluxes at various neuronal sites. By using this approach, a time-resolved image of ischemia-induced L-glutamate release in the hippocampus is shown and regional distribution of L-glutamate fluxes is discussed. Ischemia is a glucose-free hypoxic condition that is known to cause excessive L-glutamate release in the hippocampus.31 EXPERIMENTAL SECTION Materials. GluOx from Streptomyces sp. X-119-6 (5.1-6.3 units/ mg of powder)45 was obtained from Yamasa Shoyu (Choshi, Japan) and HRP from Wako Pure Chemicals Co. (Osaka, Japan). Poly(L-lysine) hydrobromide and BSA were purchased from Sigma (St. Louis, MO). Glutaraldehyde solution (10% w/w), N-(carboxymethylaminocarbonyl)-4,4′-bis(dimethylamino)diphenylamine sodium salt (DA-64), L-glutamic acid, and Triton X-100 were obtained from Wako Pure Chemicals Co. Piperazine-1,4-bis(2-ethanesulfonic acid) (PIPES) was obtained from Dojindo (Kumamoto, Japan). 2-DeoxyD-glucose was obtained from Kanto Kagaku (Tokyo, Japan). L-Glutamine, L-aspartic acid, glycine, acetylcholine, γ-aminobutyric acid (GABA), dopamine, and L-ascorbic acid were purchased from Wako Pure Chemicals Co. Other chemicals used were all of analytical reagent grade. Glass cover slips (18 × 24 mm, 0.170.25 mm in thickness) were obtained from Matsunami Glass (Tokyo, Japan) and washed in SCAT 20X (Nacalai Tesque, Inc., Kyoto, Japan) overnight. Milli-Q water (Millipore reagent water system, Bedford, MA) was used throughout the experiments. GluOx-HRP membranes were prepared according to ref 43 and stored at 4 °C. Briefly, a 5.0-µL aliquot of a poly(L-lysine) solution (5.0 µg/mL) was dropped on a cover slip, onto which another cover slip was placed. The cover slips were allowed to stand overnight for the adsorption of poly(L-lysine). Both cover slips were used for preparing GluOx-HRP membranes. A 30-µL aliquot of a GluOx-HRP solution (41 units/mL GluOx, 250 units/mL HRP, 0.050% (w/v) glutaraldehyde, 0.50% (w/v) BSA, 6.3 mM NaCl, and 6.3 mM NaH2PO4/NaOH, pH 7.4) was layered on the poly(L-lysine)-adsorbed cover slip and air-dried under darkness for 1 day. The concentration of glutaraldehyde used was lower than those (0.2-0.25%) conventionally used for immobilizing enzymes in sensor membranes.40,46 Hence, small amounts of the enzymes could diffuse into brain slices from the membrane. Solutions. A 124 mM NaCl solution containing 2.5 mM KCl, 2.0 mM CaCl2, 1.0 mM MgSO4, 26 mM NaHCO3, 1.25 mM NaH2PO4, 10 mM D-glucose, 5.3 mM DA-64, 0.05% Triton X-100, and 10 mM PIPES/NaOH, bubbled with a mixture of 95% O2/5% CO2 (final pH 7.4, abbreviated as a normal solution), was used for loading extracellular fluids in brain slices with DA-64. Triton X-100 was used for easier dissolution of DA-64. A glucose-free hypoxic solution (here called an ischemia solution) was prepared, in which D-glucose in the normal solution was replaced by an equimolar amount of 2-deoxy-D-glucose and equilibrated with 95% N2/5% CO2. A recovery solution contained 124 mM NaCl, 3.0 mM KCl, 1.3 mM MgSO4, 2.0 mM CaCl2, 26 mM NaHCO3, 1.25 mM NaH2PO4, and 15 mM D-glucose, which was saturated with a 95%O2/ 5% CO2 gas mixture. (45) Kusakabe, H.; Midorikawa, Y.; Kuninaka, A.; Yoshino, H. Agric. Biol. Chem. 1983, 47, 179-182. (46) Niwa, O.; Horiuchi, T.; Torimitsu, K. Biosens. Bioelectron. 1997, 12, 311319.

Table 1. Selectivity of the GluOx-HRP Membrane compound

responsea

400 µM L-glutamate 400 µM L-glutamate + 100 µM L-ascorbic acid 1.0 mM GABA 1.0 mM acetylcholine 1.0 mM dopamine 1.0 mM L-ascorbic acid 1.0 mM glycine 1.0 mM L-glutamine 1.0 mM L-aspartic acid

17.4 ( 1.9 15.8 ( 2.1 nrb nr nr nr nr nr nr

a Defined as R(3) -R(3) , where R(3) is the mean intensity of the o c o R component 3 min after placing an ischemia solution on the membrane and R(3)c is that in the case of an ischemia solution containing each compound. Values are mean ( SD (n ) 3). b nr, no response. The response was below three times the standard deviation of R(3)o.

Selectivity of the GluOx-HRP Membrane. The selective response of the GluOx-HRP membrane to L-glutamate over potential interfering compounds (glycine, L-glutamine, L-aspartic acid, L-ascorbic acid, GABA, acetylcholine, dopamine) was investigated by placing an ischemia solution containing a 1.0 mM concentration of each compound on the membrane (Table 1). The blocking of oxidative formation of BG by L-ascorbic acid was examined by adding L-ascorbic acid, DA-64, and L-glutamate, which were dissolved in ischemia solutions, onto the membrane in sequence. The final concentrations of L-glutamate and L-ascorbic acid were 400 and 100 µM, respectively. The presence of L-ascorbic acid had no significant effects on the response to L-glutamate. Since the tested concentration (100 µM) of L-ascorbic acid was sufficiently high for slice prepearations,40,47 the effect of L-ascorbic acid present in brain slices on color responses to L-glutamate was expected to be negligible. Capturing Digital Images. Digital images (bright-field photographs) were captured with a digital camera (Coolpix 990, Nikon, Tokyo, Japan). The JPEG images taken by the digital camera were converted into TIFF ones, which were opened with Aquacosmos version 1.3 (Hamamatsu Photonics, Hamamatsu, Japan). In the TIFF images, each pixel consisted of red (R), green (G), and blue (B) components and the intensity of each component was expressed in 256 levels (0-255). To determine which component is suitable as a digital signal for measurements of L-glutamate concentration in the enzyme membrane system, the following model experiment was performed. A 30-µL portion of a known concentration of L-glutamate dissolved in an ischemia solution was placed on a GluOx-HRP membrane on a cover slip with a micropipet. The cover slip was set on a slide glass on the stage of a stereomicroscope (SZH, Olympus, Tokyo, Japan), and digital image capturing was performed at an interval of 10 s. The captured digital images were opened on Aquacosmos. An ROI for its time course analysis was defined as the area of the solution, except for its edge. The time course analysis yielded the mean intensity (R(t), G(t), and B(t)) of each component over the ROI at a given time, t, where t is time from the spread of the ischemia solution to the capture of the digital image. Since the active region of the R component is in the range from 380 to 440 nm and from 590 to 740 nm (measured in the present study, (47) Rice, M. E.; Pe´rez-Pinzo´n, M. A.; Lee, E. J. K. J. Neurophysiol. 1994, 71, 1591-1596.

Analytical Chemistry, Vol. 75, No. 15, August 1, 2003

3777

Figure 3. (a) Schematic representation of an experimental setup for calibrating color intensity. (b) and (c) Schematic representations of model experiments used for determining the parameter ∆t for the slope analysis with L-glutamate in a standard solution (left) and expected results of the analysis (right). Figure 2. (a) Active wavelength regions of R (solid black line), G (dotted line), and B (solid gray line) components of a digital camera, measured with a Shimadzu (Kyoto, Japan) UV-240 spectrophotometer. (b) Absorption spectrum of BG formed in an ischemia solution on the GluOx-HRP membrane. The spectrum was measured according to ref 43. Concentration of L-glutamate was 500 µM. The reaction time was 2 min. (c) Relationships between L-glutamate (L-Glu) concentration in an ischemia solution placed on the GluOxHRP membrane and digital color intensity. R(t) (b), G(t) (O), and B(t) (4) at t ) 3 min were plotted. Values are mean ( SD (n ) 3).

Figure 2a) and covered the absorption maximum of BG (727 nm) (Figure 2b), R(t) decreased linearly with increasing L-glutamate concentration (Figure 2c). On the other hand, G(t) and B(t) were independent of the L-glutamate concentration. Therefore, only the R component was relevant as an analytical signal that reflects the concentration of L-glutamate. Calibrating Color Intensity. An experimental setup for calibrating the intensity of an R component with different concentrations of L-glutamate is shown in Figure 3. A square grid (area, 2.9 cm2) made of a Pt wire (diameter, 0.20 mm) was dipped in an ischemia solution and pulled up slowly. The Pt grid could retain the solution of ∼8 µL. The grid was carefully transferred on a GluOx-HRP membrane on a cover slip. The solution within 3778 Analytical Chemistry, Vol. 75, No. 15, August 1, 2003

the grid wetted the surface of the membrane. Then a 50-µL aliquot of an ischemia solution was added to the solution inside the Pt grid. The solution spread spontaneously and immediately within the Pt grid, forming a thin layer of the solution on the membrane. The thickness of the solution was estimated to be 0.2 mm from the solution volume and the surface area. The membrane was transferred onto a slide glass on the stage of a stereomicroscope, and digital images were captured at an interval of 10 s with a 7.5× objective. After 3 min, a 0.20-µL portion of an ischemia solution containing 20 mM L-glutamate (4.0 × 10-9 mol) was injected into the solution enclosed with the Pt grid, followed by capturing digital images every 10 s. The amount (moles) of L-glutamate was increased by adding repeatedly the same volume of ischemia solution containing L-glutamate at an interval of 30 s, and digital images were captured as above. The increase in the solution volume by injecting the ischemia solution was less than 4% of the initial volume. The whole procedure was repeated for different amounts (0, 8.0 × 10-9, and 1.2 × 10-8 mol) of L-glutamate. To analyze the captured digital images, an increment in the amount (moles) of L-glutamate in a unit time for each ROI rather than its concentration was used because the increment-based approach enabled us to monitor a time-dependent change in the

Figure 4. (a) Plots of R(t) vs t when the same amount of L-glutamate was repeatedly added to an ischemia solution inside a Pt grid on the GluOx-HRP membrane. The ischemia solution initially contained no L-glutamate, and a 0.20-µL portion of an ischemia solution containing (1) 0, (2) 4.0 × 10-9, (3) 8.0 × 10-9, and (4) 1.2 × 10-8 mol of L-glutamate was added at an interval of 30 s during 3.0-7.5 min. (b) Relationship between the slope of the linear part (3.0-7.5 min) of the plot (a) and the increment in amount of L-glutamate. Values are mean ( SD (n ) 3-4).

amount of L-glutamate in the ROI. At first, a ROI was defined as the entire area of the solution enclosed with a Pt grid. Then the mean intensity over the ROI was evaluated, yielding a plot of R(t) versus t (Figure 4a). The slope of the linear part (3.0-7.5 min) of the plot was calculated based on the least-squares method. The obtained slope was proportional to the increment of moles of L-glutamate in a unit time for the ROI (Figure 4b). The dynamic range was 2.8-8.3 nmol cm-2 min-1. Hence, it is a measure of the amount of L-glutamate that comes into the ROI in a unit time from the outside. Slope Analysis and Differentiation Parameters. The timedependent changes in L-glutamate fluxes in brain slices were visualized by applying a slope analysis to the captured digital images (vide infra). In the slope analysis, a regression line was prepared for a plot of R(t) versus t in the time range from (t - ∆t) to (t + ∆t) based on the least-squares method and the slope (min-1) of the regression line was calculated. To determine the time interval (∆t), two experiments were performed and the results of slope analyses with ∆t of 20, 40, and 60 s were compared. In one experiment, the amount of L-glutamate in an ischemia solution enclosed with a Pt grid was increased from 0 to 1.2 × 10-8 mol within 7 s. The other experiment was the same as that used for calibrating the intensity of an R component (vide supra). The slope analysis with ∆t of 20 s resulted in a larger noise as compared with that using ∆t ) 40 s for both types of experiments.

When ∆t of 60 s was used, the slope analysis yielded a triangleshaped plot rather than rectangle- (Figure 3b) or pulse-shaped (Figure 3c) ones. Consequently, the slope analysis with ∆t ) 40 s was the most appropriate. The response time of the enzyme membrane defined as the time required to induce 95% of the maximum response was 9 s (n ) 2) for a step change in the amount of L-glutamate from 0 to 1.2 × 10-8 mol, which was much shorter than the time width of ∆t ) 40 s. Hence, the slope analysis was not affected by the rate of the enzyme reaction. Imaging of L-Glutamate Releases in Mouse Brain Slices. Adult male ddY mice were decapitated under ether anesthesia. Coronal slices (thickness 150 µm) were cut using a microslicer (Dosaka DTK-100, Kyoto, Japan) on an ice bath. The slices were incubated for 1 h in a recovery solution at 31 °C and held at room temperature until use. The slice was spooned from the recovery solution and washed twice with a 100-µL portion of a normal solution. Then the slice was loaded with DA-64 by adding a 100-µL portion of a normal solution. After 3 min, the slice was washed three times with a 100-µL portion of an ischemia solution. Finally, a 150-µL portion of an ischemia solution was added to the slice and the slice was transferred onto a GluOx-HRP membrane. The excess solution on the slice was sucked up with a Kimwipe, so that the slice remained wet. The membrane was set on a slide glass on the stage of a stereomicroscope. Brightfield photomicrographs were taken with a digital camera every 20 s. The time required for acquisition of each image was 4 s. The magnification of the objective was usually 33×. All the measurements were done at 31 ( 2 °C. Humidity at the sample stage was maintained to be 81 ( 8% by using a plastic box in which wet sponges were placed. The duration of ischemia was defined as the time from the final addition of the ischemia solution to the capture of a digital image. Two-dimensional visual images of L-glutamate fluxes in brain slices were obtained as follows. TIFF images for the slice were divided into 12 288 ROIs. The size of each ROI was 5 × 5 pixels. The mean intensity R(t) was calculated for each ROI (25 pixels), yielding a plot of R(t) versus t. A slope analysis with ∆t of 40 s, as determined above, was applied to the plot for obtaining its slope. The same procedure was repeated for all ROIs. Then, the slopes at time t were arrayed in a work sheet of Microsoft Excel in the order of ROIs. In the work sheet, the shape of each cell was adjusted to be a square. Based on the magnitude of slopes, the color of the cell was painted green according to the color index (shown in Figure 5) in order to obtain a flux image. It should be noted that the size of ROIs and magnification of a microscope had no effect on R(t); the value of R(3 min) for a 30-µL aliquot of 400 µM L-glutamate placed on the GluOx-HRP membrane was 165.8 ( 1.0 (n ) 3) for a ROI covering the solution area under a 7.5× objective and 166.2 ( 0.7 (n ) 3) for a ROI of 5 × 5 pixels under a 33× objective. RESULTS AND DISCUSSION Time-Resolved Imaging of L-Glutamate Fluxes in Mouse Brain Slices during Ischemia. We applied the present method to time-resolved imaging of hippocampal distribution of ischemiainduced L-glutamate release in mouse brain slices. Brain ischemia is known to cause neuronal death in the hippocampus.48-50 It is thought that ischemia-induced cell death is mediated by excessive release of L-glutamate in the extracellular space and subsequent Analytical Chemistry, Vol. 75, No. 15, August 1, 2003

3779

Figure 5. Time-resolved two-dimensional visual images of L-glutamate fluxes in a mouse coronal slice during ischemia. The duration of ischemia was indicated at the left bottom of each image. In the first image, a digital monochromatic image was superimposed on the flux image. Three squares in the first image show the positions of stratum radiatum of the CA1 and stratum orien of the CA3 and DG, respectively. Lower panel shows schematic representation of a mouse brain slice. Gray region represents hippocampus. Cx, cerebral cortex; Th, thalamus.

activation of glutamate receptors in postsynaptic cells.31,49,50 Figure 5 shows the time-resolved two-dimensional visual images of L-glutamate fluxes during ischemia. The stronger the green color is, the larger the flux of L-glutamate. A weak signal representing the presence of a small L-glutamate flux was seen all over the hippocampus 5.3 min after the onset of ischemia. After ∼6 min, an intense signal originated in the stratum orien of the CA3 region and extended into the CA1 region and other parts of the CA3 region and, after 8 min, into the DG region. Then the signals in the CA1 and DG regions weakened, while an intense signal remained in the border region between CA1 and CA3. The same order of signal generation was observed from different slices (n ) 4) whose cutting position in the brain was interaural 1.501.74 mm.51 Under a normoxic condition in the presence of D-glucose (control), very weak signals were seen over the hippocampus until 8 min (Figure 6a). Then an intense signal appeared in the CA3 region and extended into the CA1 region and finally into the DG region. Although the intense signal appeared locally first in the stratum orien of the CA3 region during ischemia, such signal localization was not seen in the control slices. When GluOx was absent in the membrane, no noticeable signals were observed over the whole region of the slice even if the slice was kept for 12 min (Figure 6b). These results suggest that the signals originating in the stratum orien of the CA3 region during ischemia were actually due to L-glutamate fluxes induced by ischemia. The spatial resolution of the present method is 42 µm, which is equal to the size of a ROI (25 pixels). The effect of diffusion of the dye BG in extracellular fluids was also examined as a model (48) Kirino, T. Brain Res. 1982, 239, 57-69. (49) Lipton, P. Physiol. Rev. 1999, 79, 1431-1568. (50) White, B. C.; Sullivan, J. M.; DeGracia, D. J., O’Neil, B. J.; Neumar, R. W.; Grossman, L. I.; Rafols, J. A.; Krause, G. S. J. Neurol. Sci. 2000, 179, 1-33. (51) The cutting position of the slices in the mouse brain was estimated according to: Paxinos, G.; Franklin, K. B. J. The Mouse Brain in Stereotaxic Coordinates, 2nd ed.; Academic Press: San Diego, CA, 2001.

3780 Analytical Chemistry, Vol. 75, No. 15, August 1, 2003

experiment by adding a small aliquot ( DG (F- and t-tests, at a confidence level of 95%). No significant differences were detected between the CA1 and CA3 regions. This observation in terms of an L-glutamate flux is in accordance with the concentration-based result of in vivo microdialysis for gerbil brain under ischemia.58 However, the high level of L-glutamate release in the CA3 as well as CA1 does not explain the reported relative vulnerability; i.e., the CA1 region is highly vulnerable to ischemic injury, whereas the CA3 region is relatively resistant.32,54,55 The maximum L-glutamate flux of 2.3 ( 0.6 nmol cm-2 min-1 at the CA1 region was very close to those (0.4-4 nmol cm-2 min-1)58-60 estimated from the reported result of in vivo microdialysis under the assumption that L-glutamate comes into the dialysis tube through the full membrane area. The closeness of the magnitude of the L-glutamate flux between the two approaches (58) Mitani, A.; Andou, Y.; Kataoka, K. Neuroscience 1992, 48, 307-313. (59) Benveniste, H.; Jφrgensen, M. B.; Sandberg, M.; Christensen, T.; Hagberg, H.; Diemer, N. H. J. Cereb. Blood Flow Metab. 1989, 9, 629-639. (60) Cui, Y.; Zhang, L.; Utsunomiya, K.; Yanase, H.; Mitani, A.; Kataoka, K. Neurosci. Lett. 1999, 271, 191-194.

3782

Analytical Chemistry, Vol. 75, No. 15, August 1, 2003

confirms the reliability of the present method for visualizing L-glutamate release in terms of its flux. Ischemia-Induced L-Glutamate Release in the Mg2+Omitted Condition. It has been reported by using an enzymemodified microelectrode array that omission of Mg2+ from the extracellular solution causes L-glutamate release in all regions in rat hippocampal slices even in the normoxic condition.61 The homogeneous distribution of fluorescence signals during ischemia in the Mg2+-omitted condition has also been reported by Uchino et al., who used fura-2-loaded CHO cell lines expressing the 1/ ζ1 NMDA receptor as a detector for L-glutamate.39 We examined whether the omission of Mg2+ from an ischemia solution increases L-glutamate fluxes at various regions of mouse hippocampus. The omission of Mg2+, however, does not mean Mg2+-free, since small amounts of Mg2+ from brain slices could be present. As shown in Figure 8, an intense signal due to an L-glutamate flux was already seen over the hippocampus 4.3 min after the onset of ischemia. The intense signal continued in all regions until 8-9 min. No significant differences in the signal intensity were detected among the CA1, CA3, and DG regions. The flux of L-glutamate at 4.3 min after ischemia in the Mg2+-omitted condition was larger by 1.8-3.5-fold than that in the presence of Mg2+ for all the regions. Thus, the omission of Mg2+ accelerates L-glutamate release during ischemia. The biphasic time course observed in (61) Personal communication.

the CA1 region in the presence of Mg2+ was not clear in the Mg2+omitted condition. This is because the magnitude of L-glutamate fluxes among different slices varied to a larger extent. The results described here demonstrate the usefulness of the present approach for quantifying the magnitude of an L-glutamate flux under different physiological conditions. In contrast, the approach based on the 1/ζ1 NMDA receptor-expressing cells needs to remove extracellular Mg2+ for the receptor to work as a detector. Implication to Ischemic Injury. Although high levels of L-glutamate in the extracellular space,31,58-60,62 a region-specific intracellular Ca2+ increase,63,64 and hypoxic depolarization65,66 are well established to appear during ischemia, a direct linkage between the enhanced release of L-glutamate and the neuronal injury has not been fully established. The results obtained above show the intensity of the ischemia-induced L-glutamate flux depends on the neuronal regions of the hippocampus, and the high level of an L-glutamate flux is seen at both the highly vulnerable CA1 region and the more resistant CA3 region. The resistance of the CA3 has been related to reduced depolarization65,66 and a small increase in intracellular Ca2+ concentration.63,64 The present results suggest that the resistance of the CA3 region to ischemic injury is partly associated with the mechanism that depolarization leading to enhancement of a Ca2+ influx through NMDA receptor is not triggered even when L-glutamate is excessively released. This view is in accordance with the previous observation that the elevation of extracellular L-glutamate induced by ischemia and the subsequent activation of L-glutamate receptors occurring in the ischemic period are not the sole cause of neuronal injury.31 (62) Benveniste, H.; Drejer, J.; Schousboe, A.; Diemer, N. H. J. Neurochem. 1984, 43, 1369-1374. (63) Mitani, A.; Takeyasu, S.; Yanase, H.; Nakamura, Y.; Kataoka, K. J. Neurochem. 1994, 62, 626-634. (64) Shimazaki, K.; Nakamura, T.; Nakamura, K.; Oguro, K.; Masuzawa, T.; Kudo, Y.; Kawai, N. NeuroReport 1998, 9, 1875-1878. (65) Aitken, P. G.; Tombaugh, G. C.; Turner, D. A.; Somjen, G. G. J. Neurophysiol. 1998, 80, 1514-1521. (66) Kreisman, N. R.; Soliman, S.; Gozal, D. J. Neurophysiol. 2000, 83, 10311038.

CONCLUSION The results obtained above demonstrate that the GluOx-HRP membrane combined with a difference-image analysis provides a useful method for visualizing time-resolved regional distribution of L-glutamate release in brain slices. Although the observed signal (R(t)) corresponds to the time-integrated amount of L-glutamate release, the difference-image analysis enabled us to visualize an increase or a decrease in L-glutamate fluxes under ischemia stimulation. The digital image is captured using a conventionally used digital camera. The use of the enzyme membrane without superfusion enhances the spatial resolution of visualization of L-glutamate release to detect its regional differences within the same hippocampal slice. The difference-image analysis enables us to monitor time courses of L-glutamate release with the time resolution of 80 s. The time resolution is sufficient for monitoring L-glutamate release during ishemia, which proceeds in a time scale of several minutes. Although the GluOx-HRP membrane method was optimized for imaging regional distribution of ishcemiainduced L-glutamate release in brain slices, it is, in principle, applicable to imaging L-glutamate release at a single-cell level. The use of a high-magnification objective and a high-speed camera will improve the spatial and temporal resolution, respectively. The improvement of lower detection limit is also possible by choosing more sensitive redox dyes for the peroxidase system. The present approach is generally applicable to other neurotransmitters for which suitable enzymes are available. ACKNOWLEDGMENT This work was supported by a Grant-in-Aid for Scientific Research from the Ministry of Education, Culture, Sports, Science and Technology, Japan. Financial support from a Research Grant by College of Humanities and Sciences, Nihon University, is also acknowledged.

Received for review March 6, 2003. Accepted May 14, 2003. AC030088+

Analytical Chemistry, Vol. 75, No. 15, August 1, 2003

3783