Proc. Natl. Acad. Sci. USA Vol. 90, pp. 7661-7665, August 1993 Neurobiology Photostimulation using caged glutamate reveals functional circuitry in living brain slices (neocortex/whole- recording/intrinsic circuits) EDWARD M. CALLAWAY* AND LAWRENCE C. KATZt Department of Neurobiology, Box 3209, Duke University Medical Center, Durham, NC 27710 Communicated by Dale Purves, May 6, 1993

ABSTRACT An approach for high-spatial-resolution map- stimulated axons of passage along with cell bodies did not ping of functional circuitry in living mammalian brain slices limit its utility. However, mapping circuitry in the dense has been developed. The locations of neurons making func- neuropil of the mammalian central nervous system necessi- tional synaptic connections to a single neuron are revealed by tates a method that does not stimulate axons of passage. In photostimulation of highly restricted areas of the slice (50-100 addition, the method ofFarber and Grinvald (7) was based on ,um in diameter) while maintaining a whole-cell recording of the light-induced formation of small transient pores in the the neuron ofinterest. Photostimulation is achieved by bathing neuronal membrane and, thus, could only be used to activate brain slices in a molecularly caged form ofthe neurotransmitter a particular neuron a few times before it was killed by glutamate [L- a-(4,5-dimethoxy-2-nitrobenzyl) phototoxic damage. Furthermore, the performance of their ester], which is then converted to the active form by briefpulses photostimulation probes varied between cell types and was (<1 ms in duration) ofultraviolet irradiation. Direct activation occasionally species-specific. Here we report development ofreceptors on recorded neurons in rat hippocampus and ferret of a photostimulation method, based on the use of caged visual cortex demonstrates that photostimulation is reliable and glutamate (8), designed to achieve high-resolution mapping of reproducible and can be repeated at the same site at least 30 functional microcircuitry in living mammalian brain slices. times without obvious decrement in neuronal responsiveness. This approach is based on the fact that virtually all neurons Photostimulation of presynaptic neurons at sites distant to the in the mammalian central nervous system can be activated by recorded neuron evoked synaptic responses in hippocampal the excitatory amino acid neurotransmitter glutamate. By and cortical cells at distances of up to several millmeters from combining photostimulation with whole-cell patch-clamp re- the recorded neuron. Stimulation of 25-100 distinct presyn- cordings, the regions within a brain slice that contain neurons aptic sites while recording from a single postsynaptic neuron making functional connections to a single neuron can be was easily achieved. Caged glutamate-based photostimulation mapped with high resolution and the strengths of the con- eliminates artifacts and limitations inherent in conventional nections can be measured. stimulation methods, including stimulation ofaxons ofpassage, desensitization, and poor temporal resolution of "puffer" pipettes, and current artifacts of iontophoretic application. MATERIALS AND METHODS This approach allows detailed physiological investigation and Overview. Brain slices were prepared by conventional manipulation of the complex intrinsic circuitry of the mam- methods and submerged in artificial cerebral spinal fluid malian brain. (ACSF) containing 1 mM "caged glutamate" [L-glutamic acid a-(4,5-dimethoxy-2-nitrobenzyl) ester]. The synthesis of this Understanding the detailed organization of neuronal net- compound and some of its properties, along with several works is important for elucidating computations responsible other amino acid neurotransmitters, were described by for brain function. The limitations of presently available Wilcox et al. (8). Whole-cell recordings were obtained from methods for revealing such circuits are especially apparent in neurons in selected regions of brain slices and maintained the study of detailed local circuits, as in the mammalian while stimulating small groups of neurons at selected neigh- cerebral cortex, in which very closely spaced neurons often boring locations within the slice. Stimulation was achieved by make up distinctly different components of the network (1). converting caged glutamate to active glutamate by using an Conventional physiological and anatomical methods, such as intense 800-,sec-duration flash of ultraviolet irradiation fo- current source density analysis (2, 3), single-unit recordings, cused through the epifluorescence optical pathway of an cross-correlation analyses, intracellular recording, and dye inverted microscope. The inverted microscope was mounted filling (4-6), allow only limited insight into the connectivity on an x-y stage, allowing it to be moved freely beneath the of neuronal assemblies. To overcome these limitations, Far- fixed stage containing the brain slice without disrupting the ber and Grinvald (7) developed an approach based on light- whole-cell recording. Repeated stimulation at closely spaced based neuronal stimulation ("photostimulation") for dissect- locations resulted in high-resolution functional mapping of ing neuronal networks. They designed photostimulation mol- the local circuitry impinging on the recorded neuron. A ecules (analogs of voltage-sensitive dyes) that allowed detailed description of the technique is provided below. stained neuronal elements to be activated by small spots of Slice Preparation and Whole-Cell Recording. Brain slices light and were able to sequentially stimulate numerous were prepared from the hippocampus, somatosensory cor- neurons in invertebrate ganglionic preparations to identify tex, or visual cortex of rats aged from 2 weeks postnatal to those that were presynaptic to an intracellularly impaled neuron. Since single neurons could easily be targeted for Abbreviations: CNQX, 6-cyano-7-nitroquinoxaline-2,3-dione; stimulation in the isolated ganglia, the fact that their method AMPA, a-amino-3-hydroxy-5-methyl-4-isoxazole propionate; TTX, tetrodotoxin; EPSC, excitatory postsynaptic current. *Present address: Department of Physiology and Neuroscience The publication costs of this article were defrayed in part by page charge Program, University ofColorado School ofMedicine, Campus Box payment. This article must therefore be hereby marked "advertisement" B138, 4200 East Ninth Avenue, Denver, CO 80262. in accordance with 18 U.S.C. §1734 solely to indicate this fact. tTo whom reprint requests should be addressed. 7661 Downloaded by guest on October 1, 2021 7662 Neurobiology: Callaway and Katz Proc. Natl. Acad. Sci. USA 90 (1993) adult, from the visual cortex of ferrets aged 4-5 weeks laser alignment spot on each image and then superimposed to postnatal, or from the visual cortex of adult cats. Slices (400 create a single map. ,um thick) were cut using a specialized slicer and maintained at 33-34°C in an interface holding chamber as described (1). RESULTS At least 1 h after cutting, individual slices were moved to a recording chamber on the fixed stage of an inverted micro- Direct Neuronal Stimulation. For this photostimulation scope and submerged in ACSF containing 1 mM L-glutamic method to be useful for mapping circuitry, uncaged glutamate acid a-(4,5-dimethoxy-2-nitrobenzyl) ester [caged glutamate must be capable of stimulating neurons sufficiently to pro- (8), obtained from Molecular Probes] at room temperature. In duce an . We used whole-cell recordings in a some control experiments, the ACSF contained no caged voltage-clamp mode (holding potential, -66 mV) to monitor glutamate. The slice was held against the coverslip on the currents evoked by light flashes positioned directly under the bottom of the chamber with small pieces of silver wire placed recorded neurons. Large inward currents were invariably over its ends. About 5 ml of ACSF was continously reper- measured and were large enough to overcome the voltage fused through the recording chamber with a peristaltic pump clamp, resulting in regenerative action potentials (Fig. 1). (at -0.5 ml/min) and reoxygenated with 95% 02/5% CO2 by Inward currents were evoked in all 30 cells tested including bubbling in a 5-ml syringe used as a secondary reservoir. 5 in rat hippocampus, 4 in rat cerebral cortex, 13 in ferret Whole-cell recordings were obtained in selected regions of visual cortex, and 8 in cat visual cortex. No decrement in brain slices as described by Blanton et al. (9). The electrodes responsiveness was observed with repeated stimulation of used had resistances of 5-10 Mfl and were filled with the neurons, which were in some cases activated by as many as following solution: 130 mM cesium gluconate/10 mM 30 flashes. Responses were only observed when the objective CsCl2/10 mM Hepes/li mM EGTA/1 mM CaCl2/1 mM was focused into the slice itself. Focusing the objective even MgCl2/2 mM ATP/0.3 mM GTP, pH 7.4. After whole-cell slightly below the slice greatly attenuated the responses. In recordings were obtained, neurons were voltage-clamped at several cases we noted that relatively small changes in focus -66 mV and the whole-cell currents evoked by "photostim- (10-50 ,um) produced significant changes in the magnitude of ulation" were recorded using an Axopatch 1-D amplifier response, implying that the technique can provide at least (Axon Instruments, Burlingame, CA) and saved using some z-axis resolution, especially when high-numerical- PCLAMP software (Axon Instruments). aperture objectives are used. Uncaging Protocol. Glutamate was uncaged in small regions These responses clearly resulted from the light-induced of the slice by a flash of light focused through the optics of an release of glutamate since responses were not evoked in inverted epifluorescence microscope (Zeiss IM 35). A xenon slices containing no caged glutamate (no detectable inward flashlamp was placed in the epifluorescence lamp housing of current from 23 flashes in 5 cells) or when a shutter in the the microscope (equipped with a quartz collector) and was epifluorescence light path was closed to prevent the light powered by a Chadwick-Helmuth power supply (model 238). from reaching the specimen (9 cells, Fig. 1). In other exper- The power supply was typically charged maximally to 250 J iments (2 cells), addition ofthe glutamate antagonist 6-cyano- and discharged to deliver a 800-,usec-duration flash (resulting 7-nitroquinoxaline-2,3-dione (CNQX) at 20 uM reduced the in delivery of =200 mJ). The light flash was focused through magnitude of light-induced responses by -40%; subsequent the normal epifluorescence light path, which included a heat addition of 5-amino-phosphovaleric acid at 100 ,uM com- filter for removing wavelengths above 700 nm. The filter cube pletely blocked responses. These observations indicate that housed only a dichroic mirror that reflected all wavelengths light-evoked responses resulted directly from the uncaging of <510 nm to the specimen. The light was focused to a small glutamate and its activation of glutamate receptors and not spot with a x63 Zeiss Neofluor oil-immersion objective (1.3 from other photolysis products. The observation that CNQX numerical aperture), which was stopped down to a radius of reduces the light-evoked response suggests that free gluta- 50 uni with an adjustable aperture in the light path. The mate impurities that may be present in the caged glutamate do objective was focused to a depth of -200 ,um into the slice by not result in inactivation of a-amino-3-hydroxy-5-methyl-4- visualizing the tissue through the normal optical pathway of isoxazole propionate (AMPA) receptors by desensitization the microscope. (see Discussion). The precise position of the microscope objective under the In preliminary experiments using conventional intracellu- slice (location of photostimulation) was determined prior to lar recording with high-resistance (50-200 Mfl) microelec- discharge of light flashes using a HeNe laser alignment trodes, responses could be elicited by the light flash alone (in system and recorded by taking a video image with a charge- the absence of caged glutamate) when the light directly coupled device camera attached to a phototube on a dissect- impinged on the recording electrode. It appears that intense ing microscope (Wild M5; magnification, x 120 or x 250) light changes the resistance of high-resistance sharp elec- placed above the preparation. The long-wavelength light from the HeNe laser was aimed into the photoport of the inverted microscope where the optics directed it through the dichroic mirror and it was focused to a small spot (<50 gm) 0 by the microscope objective. The laser was aligned so that the spot was at the center of the field. The long-wavelength (pA) alignment light (632 nm) is not absorbed by the caged -500 glutamate (8), produces no photolysis, and therefore, cannot "uncage" the glutamate. After aligning the microscope objective under a desired -1i000 photostimulation location, the flashlamp was discharged and 0 25 50 75 the signal from the patch-clamp amplifier for the subsequent Time (msec) msec was saved to the computer. Maps of response 200-400 FIG. 1. Response of a hippocampal pyramidal cell in CAl to strength corresponding to various stimulus locations (see direct photostimulation centered on the cell body. In the lower trace, Figs. 3 and 4) were constructed by importing the video images photostimulation resulted in large inward currents with essentially no indicating stimulus locations to an Apple IIfx computer. latency, which produced regenerative spikes within 10 msec as the Circles either color-coded or size-coded to indicate peak cell escaped from voltage clamp. The upper trace shows a shutter- inward currents after stimulation were "pasted" over the closed control; no inward currents were produced. Downloaded by guest on October 1, 2021 Neurobiology: CallawayNeurobiology:and Katz Callaway and Katz Proc. NatiL Acad. Sci. USA 90 (1993) 7663 trodes, resulting in a transient changes in applied holding currents that give the illusion of neuronal responsiveness. Similar effects can be induced by laser stimulation during conventional intracellular recording (7). Such effects were not observed during whole-cell recording, under either volt- age or current clamp, presumably because of the relatively low resistance of the electrodes used (5-10 Mfl) and the low applied currents necessary to maintain steady membrane polarization in the resting state (typically <0.1 pA). Spatial Resolution of Photostimulation. To determine whether the effects of uncaged glutamate were spatially restricted to the region of light flashes, slices were bathed in tetrodotoxin (TTX) to eliminate synaptic responses that might be evoked by glutamate release distant from the recorded neuron. Closely spaced light flashes were then targeted to various locations around the cell body to map its responsiveness, as illustrated in Fig. 2. The recorded neuron was in layer 2/3 of a slice from ferret primary visual cortex. B0 Stimulation centered on the cell body (Fig. 2B) evoked a large 1 prolonged inward current with a time course suggestive of a strong contribution from voltage-dependent chan- nels. This response lacked the fast-inactivating component observed in the absence of TTX (Fig. 3B), which was 0- -600 presumably mediated by voltage-dependent sodium chan- nels. The color-coded spots in Fig. 2A show the peak magnitudes of inward currents evoked by light flashes cen- tered at the spot locations. Two ofthese responses (responses 2 and 3) correspond to the current traces shown in Fig. 2C, which were evoked by stimulation at 75 and 125 ji&m lateral to -1200 the cell body, respectively. A clear inward current was evoked by stimulation at a distance of 75 Am but not 125 am. Stimulation at locations closer to the pial surface evoked C 0 3 responses from further distances (Fig. 2A), suggesting that the recorded neuron had an apical dendrite extending to the 2 pial surface; this is the morphology of the overwhelmin majority of pyramidal neurons in layer 2/3 of primary visual CL -50 cortex. Based on the extent of dendritic processes in typical layer 2/3 neurons in ferret primary visual cortex, it appeared that the neuron only responded when a portion of the light flash overlapped with part ofthe dendritic arbor. Stimulation just above the pial surface of a slice (in the surrounding. medium) induced no response in a layer 2/3 neuron whereas 0 50 100 150 stimulation only 100 um below the pia evoked a clear inward current. In other cases, responses were evoked by stimula- Time (msec) tion at the border between the white matter the and cortical FIG. 2. (A) Responses of a cortical neuron in the presence of plate but not by stimulation confined to the white matterjust TTX. The dotted line indicates the pial surface; the arrow shows the 100jum away. This observation also confirms that axons of position of the cell body. The recording pipette is visible to the left passage are not activated by photostimulation. of the arrow. Responses of a single cell, when all synaptic commu- Synaptic Responses and Mapping Connections. Mapping nication is blocked by TTX, are coded by location and color. The experiments in the absence of TTX revealed synaptic re- numbers correspond to their respective traces in B and C. Responses sponses evoked by photostimulation at distances of up to are confined to a restricted region corresponding to the approximate several millimeters from the cell bodies of recorded neurons dimnensions of the cells' dendritic field. No responses were observed 3 and when stimulation was just outside the slice or >100 jan lateral to the (Fig. 4). cell body. The sharp boundaries between responsive and nonrespon- Fig. 3A illustrates responses from a neuron in the CAl sive zones indicate that is restricted. of rat photostimulation spatially (Bar region hippocampus after photostimulation of distant = 100 pLm.) (B and C) Inward currents (traces 1-3) evoked by presynaptic neurons. Photostimulation in the CA3 region ==1 photostimulation at the positions indicated by the numbers in A. mm from the recorded neuron elicited an inward current of60 Photostimulation at position 1 resulted in a large inward current with pA (lower trace). This response is attributed to the Shaeffer the characteristic of a regenerative calcium current. Stimulation at collateral pathway. Stimulation in the dentate gymus region position 2 produced a much smaller relatively slow inward current ~-4 mm from the recorded neuron resulted in a smaller and peaking at ='50 msec. Note the absence of excitatory postsynaptic slower response, with a peak inward current of 15 pA. These currents (EPSCs). Stimulation at position 3, 125 p.m from the cell results demonstrate responsiveness at distances that clearly body, resulted in no detectable response. do not result from direct stimulation of the recorded neuron. This is also demonstrated in the experiments illustrated in inward current resulting in an action potential followed by a Fig. 3 B and C and 4. In these experiments, responses of prolonged inward current (Fig. 3B). Photostimulation at more neurons in layer 2/3 of ferret visual cortex were measured lateral locations elicited smaller inward currents with the a-ft-er photostimulation at numerous sites within layer 2/3 shape and time course of EPSCs. Stimulation at distances of lateral to the recorded neurons. The traces in Fig. 3 B and C 200, 400, and 600 pLm evoked peak inward currents of 250, correspond to positions 0, 2,4, and 6 of the response map in 125, and 30 pA, respectively (traces in Fig. 3C, bottom to Fig. 4A. Stimulation at the cell body (position 0) elicited a fast top). Fig. 4B illustrates a similar map in which responses were Downloaded by guest on October 1, 2021 7664 Neurobiology: Callaway and Katz Proc. Natl. Acad ScL USA 90 (1993) 50 A 0

-50

-100 B 0

< -25000

-50001 C

0 50 100 150 Time (msec) FiG. 3. (A) Synaptic response of a CAl pyramidal cell after photostimulation of presynaptic neurons. The lower trace shows prominent inward currents, with a latency of "10 msec, induced by FiG. 4. Maps ofphotostimulation-induced responses in layer 2/3 photostimulation in CA3. The upper trace shows the response of ferret visual cortex. In these maps the area of each circle elicited by stimulation in the dentate gyrus, 4 mm from the recorded corresponds to the magnitude of the peak response. (A) Map of 14 cell. A small (15 pA) excitatory postsynaptic current was reliably stimulation sites lateral to a cortical neuron. Synaptically evoked induced at a latency of -'40 msec. (B and C) Photostimulation responses were detected up to 1.1 mm lateral to the recorded cell. mapping of a neuron in layer 2/3 of ferret visual cortex. (B) (Bar = 100 pm.) (B) Map ofresponses evoked by 25 photostimulation Photostimulation was centered on the cell body, resulting in a large sites with the same conventions as in A. Note that the magnitude of regenerative inward current with a latency of "'25 msec. This the response first declines and then increases with increasing lateral corresponds to the position indicated by the arrow in Fig. 4A. distance. This may correspond to the clustered horizontal connec- Stimulation at more lateral locations in layer 2/3 (C) produced tions prominent in this region of cortex. (Bar = 100 pm.) EPSCs of different magnitudes and latencies. The bottom trace results from stimulation '200 pm lateral to the cell body; the middle logical features (i.e., dendritic morphology and axonal pro- trace is 400 pm lateral, and the upper trace is 800 ,um lateral. Note jection pattern as revealed by intracellular staining) of a the presence ofmultiple EPSCs in each trace and increases in latency single neuron to be understood in the context of the sources to the onset and peak of each trace with increasing lateral distance. of the numerous inputs that impinge on it rather than in isolation. Also, the resolution of the approach allows input measured after photostimulation at 25 locations lateral to the sources to be identified at a level that is currently impossible recorded neuron. Note that in the experiments ofFig. 4, clear with neuronal tracers. inward currents were evoked by photostimulation at dis- The use of "caged" glutamate and flash photolysis pre- tances of up to 1.1 mm. This is considerably farther than the sented here is an extension of previous work using caged distance at which responses could be evoked in TTX-treated compounds (for example Ca2+ and GTP) to control intracel- slices (Fig. 2) and we therefore conclude that the long- lular signaling (for review, see refs. 10-12). Caged (or pho- distance responses are synaptic, resulting from the direct toisomerable) neurotransmitters or neurotransmitter analogs stimulation of neurons presynaptic to the recorded neuron. can be used to activate machinery used for intercellular communication. For example, cis-3,3'-bis[a-(trimethylam- DISCUSSION monium)methyl]azobenzene dibromide and caged carbachol have been used to investigate the activation of acetylcholine Despite extensive study, understanding ofthe organization of receptors (10, 13-14). In principle the other caged neuro- functional local circuits in the central nervous system ofadult transmitters, including glutamate, glycine, aspartate, and and developing mammals remains limited. Most techniques, 'y-aminobutyric acid (8), could also be used for studies ofthis like intracellular staining and single-unit recording, provide nature. We have not studied the kinetics of the photochem- information about only a single component of a neuronal ical reaction using this compound. Measurements by others network; relationships between the individual components have determined that the reaction requires -100 ,sec, which must be inferred. Anatomical methods provide little infor- is somewhat long for kinetic measurements but adequate for mation about functional connectivity. We have developed a our purposes. The actual concentration of free glutamate method for mapping neural circuitry in living brain slices that achieved in the slice is also difficult to estimate, although we is based on the premise of Farber and Grinvald (7) that more surmise that it must be in the range of 50-100 ,uM to elicit detailed information about the organization of functional action potentials. neural circuits can be obtained by determining the sources of Although we describe a method for mapping the locations functional input to a single neuron. This allows the morpho- of neurons connecting to a postsynaptic cell, this photostim- Downloaded by guest on October 1, 2021 Neurobiology: CaUaway and Katz Proc. Natl. Acad. Sci. USA 90 (1993) 7665 ulation technique could also be used for a number of other flashlamp to evoke glutamate release eliminates electrical types of experiments. For example, the integration of excit- artifacts associated with discharge of the flashlamp, greatly atory signals impinging on different portions of a neuron's reduces the time necessary to produce maps, and produces a dendritic arbor has been the subject of extensive theoretical more intense and better focused light beam (M. B. Dalva and study (15-18); but there is very little experimental data L.C.K., unpublished observations). The ability to focus laser relating to the hypotheses that have been generated. In light to a much smaller spot (in three dimensions) should theory, spatially restricted release of caged glutamate by a improve spatial resolution significantly, allowing mapping on very small focused beam of laser light could be targeted to a significantly finer scale. specific portions ofthe dendritic arbor of a neuron filled with fluorescent dye and electrically monitored by whole-cell L.C.K. is an L. P. Markey Scholar. This work was supported in part by agrant from the L. P. Markey Charitable Trust. This research recording. Given a means of quickly relocating the light was also supported by National Institutes of Health Grants EY06128 beam, various spatial and temporal patterns of dendritic (E.M.C.) and EY07960 (L.C.K.). activation could be achieved. We have recently added bio- cytin, an intracellular marker, to our whole-cell electrodes. 1. Katz, L. C. (1987) J. Neurosci. 7, 1223-1249. This results in intracellularly labeled cells whose morphology 2. Mitzdorf, U. & Singer, W. (1978) Exp. Brain Res. 33, 371-394. can be directly related to the pattern of photic responses. 3. Friauf, E. & Shatz, C. J. (1991)J. Neurophysiol. 66, 2059-2071. Despite the success of the basic photostimulation and 4. Gilbert, C. D. & Wiesel, T. N. (1979) Nature (London) 280, "mapping" approaches reported here, this methodology has 120-125. 5. Gilbert, C. D. & Wiesel, T. N. (1983) J. Neurosci. 3, 1116- several potential limitations. The use ofslice preparations has 1133. the inherent limitation that circuitry can only be studied along 6. Martin, K. A. C. & Whitteridge, D. (1984) J. Physiol. 353, two axes; we also do not believe that photostimulation is 463-504. confined to the plane of focus of the flashlamp. It may also 7. Farber, I. C. & Grinvald, A. (1983) Science 22, 1025-1027. be necessary to use the inverted microscope configuration to 8. Wilcox, M., Viola, R. W., Johnson, K. W., Bilhington, A. P., prevent the release ofcaged glutamate in the large volume of Carpenter, B. K., McCray, J. A., Guzikowski, A. P. & Hess, solution superfusing the slice. There may also be limitations G. P. (1990) J. Org. Chem. 55, 1585-1589. 9. Blanton, M. G., Lo Turco, J. J. & Kriegstein, A. R. (1989) J. on the temporal resolution ofthis technique ifglutamate acts Neurosci. Methods 30, 203-210. for long periods once released. The effective period of 10. Gurney, A. M. & Lester, H. A. (1987) Physiol. Rev. 67, activation after glutamate release is probably very short 583-617. however, due to uptake by glial cells (19). There may also be 11. McCray, J. A. & Trentham, D. R. (1989) Annu. Rev. Biophys. toxic effects of side products of the photolysis of caged Biophys. Chem. 18, 239-270. glutamate (for review, see ref. 12) but we have seen no 12. Kaplan, J. H. & Somlyo, A. P. (1989) Trends Neurosci. 12, evidence of toxicity with repeated stimulation. 54-59. A final concern is that glutamate impurities in the solution 13. Milburn, T., Matsubara, N., Billington, A. P., Udgaonkar, of caged glutamate may elevate resting glutamate levels J. B., Walker, J. W., Carpenter, B. K., Webb, W. W., Marque, J., Denk, W., McCray, J. A. & Hess, G. P. (1989) leading to desensitization. This would likely have Biochemistry 28, 49-55. differential effects on activation of neurons using different 14. Gorne-Tschelnokow, U., Hucho, F., Naumann, D., Barth, A. types of glutamate receptors with various susceptibilities to & Mantele, W. (1992) FEBS Lett. 309, 213-217. desensitization [i.e., N-methyl-D-aspartate vs. AMPA recep- 15. Rall, W. (1967) J. Neurophysiol. 30, 1138-1168. tors (20-21)]. The observation that light-evoked responses 16. Lev-Tov, A., Miller, J. P., Burke, R. E. & Rall, W. (1983) J. are diminished by the addition ofthe AMPA receptor blocker Neurophysiol. 50, 399-412. CNQX indicates, however, that at least some AMPA recep- 17. Shepherd, G. M., Brayton, R. K., Miller, J. P., Segev, I., tors remain active in the presence ofcaged glutamate. We feel Rinzel, J. & Rall, W. (1985) Proc. Natl. Acad. Sci. USA 82, that it is that neurons are from 2192-2195. likely protected glutamate 18. Koch, C., Douglas, R. & Wehmeier, U. (1990) J. Neurosci. 10, impurities due to uptake by glial cells in the slice preparation; 1728-1744. in a preparation of isolated or cultured neurons such impu- 19. Hertz, L. (1979) Prog. Neurobiol. 13, 277-323. rities could present major problems. We envision a number 20. Trussel, L. O., Thio, L. L., Zorumski, C. F. & Fischbach, of improvements to this technique that are expected to G. D. (1988) Proc. Natl. Acad. Sci. USA 85, 4562-4566. enhance its capacity to reveal local brain circuits. The use of 21. Zorumski, C. F. & Thio, L. L. (1992) Prog. Neurobiol. 39, a laser light source (such as a UV-argon laser) rather than a 295-336. Downloaded by guest on October 1, 2021