標(biāo) 題: Nature:加州大學(xué)蒲慕明實(shí)驗(yàn)室移動(dòng)的光調(diào)節(jié)頂蓋神經(jīng)元敏感性 Nature 419, 470 - 475 (2002); doi:10.1038/nature00988 Moving visual stimuli rapidly induce direction sensitivity of developing tectal neurons FLORIAN ENGERT*??, HUIZHONG W. TAO*?, LI I. ZHANG?§ & M U-MING POO* * Division of NeuroBioLogy, Department of molecular and Cell Biology, University of California, Berkeley, California 94720, USA § Keck Center of Integrative Neuroscience, University of California, San Franci sco, California 94143, USA ? These authors contributed equally to this work ? Present address: Department of Molecular and Cell Biology, Harvard Unive rsity, Cambridge, Massachusetts 02138, USA. Correspondence and requests for materials should be addressed to M.-m.P. (e-mail : mpoo@uclink.berkeley.edu). During development of the visual system, the pattern of visual inputs may have a n instructive role in refining developing neural circuits1-4. How visual inputs of specific spatiotemporal patterns shape the circuit development remains largel y unknown. We report here that, in the developing Xenopus retinotectal system, t he receptive field of tectal neurons can be 'trained' to become direction-sensit ive within minutes after repetitive exposure of the retina to moving bars in a p articular direction. The induction of direction-sensitivity depends on the speed of the moving bar, can not be induced by random visual stimuli, and is accompan ied by an asymmetric modification of the tectal neuron's receptive field. Furthe rmore, such training-induced changes require spiking of the tectal neuron and ac tivation of a NMDA (N-methyl-D-aspartate) subtype of glutamate receptors during training, and are attributable to an activity-induced enhancement of glutamate-m ediated inputs. Thus, developing neural circuits can be modified rapidly and spe cifically by visual inputs of defined spatiotemporal patterns, in a manner consi stent with predictions based on spike-time-dependent synaptic modification. Spontaneous and experience-evoked activities in the developing brain can influen ce the refinement of developing nerve connections into mature neural circuits. I n the visual system, rearing kittens with an artificial squint leads to failure in the development of binocular response properties of striate cortex neurons5. Blockade of spontaneous waves of retinal activity also disrupts eye-specific seg regation of retinal inputs to the lateral geniculate nucleus6, 7. Synchronizing retinal inputs by strobe light or electrical stimulation affects formation of no rmal receptive field properties in various systems8-10. Furthermore, an instruct ive role of visual inputs is indicated by the appearance of visual modules in th e auditory cortex of the ferret after rewiring of retinal inputs11, 12. In the p resent study, we found a rapid and specific modification of receptive field prop erties of tectal neurons by the visual input of a defined spatiotemporal pattern , in a manner consistent with hebbian synaptic modification as a mechanism for a ctivity-dependent changes in visual circuits13, 14. The effect of visual experience on the receptive field properties of tectal neur ons was examined in developing Xenopus tadpoles. Patterned visual inputs were us ed to stimulate the retina, and tectal cell responses were monitored with in viv o perforated whole-cell recording methods (Fig. 1a). First, we mapped the recept ive field of the tectal neuron by random and sequential flashing of a white squa re at various locations on the retina (see Methods). The integrated charge of st imulus-evoked compound synaptic currents (CSCs) was measured within a defined wi ndow for the more prominent 'off' responses (Fig. 1b). The measured value at eac h location was represented in grey scale as one element of an 8 7 grid that cov ered the total area of the projected visual image on the retina (Fig. 1c). This analysis based on CSCs reveals a large receptive field (50–80% of the retina, n = 12) of tectal neurons at these early stages (42?C45), consistent with a diffuse retinotectal connectivity during early developme nt15, 16. Figure 1 Mapping the receptive field of developing Xenopus tectal neurons. Fu ll legend High resolution image and legend (36k) To examine the effect of patterned visual inputs on the receptive field property of tectal neurons, we stimulated the retina with white moving bars (20-μm wide, speed 0.3 μm ms-1) in four orthogonal directions (right, down, left and up, Fig. 2a) and recorded the responses of tectal cells evoked by moving vi sual stimuli. In all tectal cell responses (voltage-clamped at -70 mV, n = 20), we found no apparent directional preference to moving bars in any of the four di rections, in terms of the total charge of CSCs evoked by each moving stimulus (s ee Fig. 2b, green bars), although these CSCs exhibited a distinct profile for ea ch direction (Fig. 2a). As these developing tectal neurons had no apparent direc tional preference, we inquired whether repeated exposure of directional stimuli can 'train' these neurons to acquire direction-sensitivity in their responses. A fter the initial assay of tectal cell responses to moving-bar stimuli in each of the four directions (voltage-clamped at -70 mV), the recording was switched to current clamp and the retina was exposed to 60 sweeps of the moving bar (speed 0 .3 μm ms-1, frequency 0.2 Hz) in one specific, but randomly chosen, direct ion (rightward in Fig. 2). In many neurons, each sweep elicited reliably a train of spikes (Fig. 2b, inset). When the tectal cell responses were assayed again a fter the training, we found that the response to the training stimulus (bar to t he right in Fig. 2a) was enhanced to a level significantly larger than those for other directions. In Fig. 2a, the effect is further illustrated by superimposin g the current traces before and after training for each direction. The enhanceme nt of CSCs also led to an increased spike rate of the tectal neuron in response to the training stimulus (Fig. 2c). In most experiments, we monitored CSCs (in v oltage clamp) instead of spiking activities to avoid spiking-dependent changes c aused by testing. Figure 2d depicts the time course of training-induced changes in seven experiments, for which the training stimulus reliably induced spiking o f the tectal cell. The enhancement of tectal cell response to the test stimulus of the trained direction was persistent for up to 50 min. Figure 2 Selective enhancement of tectal cell responses by training with a movi ng bar. Full legend High resolution image and legend (127k) In many regions of the developing brain14, 17-20, the timing of spiking in pre- and postsynaptic neurons is critical for the induction of long-term potentiation (LTP) and long-term depression (LTD). Modelling studies indicate that such spik e-time-dependent plasticity may provide a mechanism for the development of direc tion-selectivity of visual circuits for detecting moving stimuli21-24. Moving ba rs probably evoke consecutive spiking of neighbouring retinal ganglion cells and of the tectal cells they innervate, with specific temporal relationships. Such a spatiotemporal pattern of spiking may be essential for triggering synaptic cha nges underlying the development of direction-selectivity. We further tested this idea by using bar stimuli that moved at slower speeds. As shown in Fig. 3a, no significant direction-specific change was induced by training with the slow-movi ng bar (0.1 μm ms-1), whereas bars moving with a medium speed (0.2 μ m ms-1) induced a slight increase in the tectal response to the training stimulu s. Comparing the patterns of tectal spikes, we found that the fast bar generated spikes with significantly shorter inter-spike intervals (Fig. 3d), although the total number of spikes evoked by each sweep was the same (5.4 0.8 and 5.6 0.7 for the fast and slow sweeps, respectively). Thus the timing of sequential exci tation of retinal cells seems to be critical for inducing direction-specific cha nges in the tectal neurons. In addition, we found that after training with the f ast-moving bar, the tectal response became enhanced only for the test bar that m oved at the fast but not the slow speed (Fig. 3b). Thus the circuit modification depends on the speed of the training stimulus, and the modified circuit is spec ific for the detection of the training stimulus. Figure 3 Effect of training with stimuli of different spatiotemporal patterns. Full legend High resolution image and legend (52k) The importance of the spatiotemporal pattern of retinal excitation in the observ ed training effects was further examined by exposing the retina to randomized tr aining stimuli. We used non-overlapping white squares that flashed in random seq uence and covered the same area of the visual field for the same amount of time as the fast-moving bar stimulus (Fig. 3c). Each set of random flashes evoked a s imilar total number of spikes (5.3 0.7) as that by a sweep of the fast-moving b ar, but the distribution of inter-spike intervals was significantly different, w ith spikes dispersed over longer intervals (Fig. 3d). After training with 120 se ts of random flash stimuli, we found no enhancement of tectal responses in any d irection (Fig. 3c). Thus the pattern of spiking, presumably in both pre- and pos tsynaptic neurons, is critical in the observed induction of direction-sensitive responses. Finally, we tested the effect of training with bars of four orthogona l directions, with a random sequence of occurrence but the same number (60) of s weeps for each direction. We found that the tectal responses after the training were enhanced in all four directions (Fig. 3c), although the level of enhancemen t was lower than that induced by the unidirectional training. Thus the tectal ce ll is not pre-determined to be direction-selective in a particular direction. Th e appearance of direction-selectivity in vivo presumably requires an imbalance o f moving visual stimuli and other mechanisms for consolidating the acquired resp onse properties of the tectum. To explore cellular changes induced by the training stimuli, we examined further the receptive fields of tectal cells before and after training. To facilitate r apid mapping, we used a set of ten vertical and ten horizontal flashing bars in random sequence to determine the receptive field profiles of the tectal neuron i n horizontal and vertical directions, respectively (Fig. 4a). After training wit h fast bars in one direction, an asymmetric change in the profile of the tectal receptive field was revealed by the enhanced tectal cell responses only for upst ream test bar locations (Fig. 4a). As summarized in Fig. 4b (top panel), asymmet ric receptive field modification was induced by fast-moving bars along the direc tion of the movement, whereas no apparent asymmetry was found for receptive fiel d in the perpendicular dimension. This is analogous to the asymmetric change in hippocampal place field induced by unidirectional movement of a rat in a closed track25, which is consistent with the prediction by a computational model based on asymmetric spike-time-dependent synaptic modification26. Spiking of tectal ne urons evoked by the fast-moving bar appears with a delay of a few hundred ms aft er the onset of the stimulus sweep on the projection screen (Fig. 4c), with most spikes clustered during the early phase of the sweep, where modification of the receptive field was most prominent. In addition, consistent with the involvemen t of spike-time-dependent plasticity, no change in the receptive field profile w as observed when the tectal cell was hyperpolarized to be prevented from spiking during training (Fig. 4b, middle panel), or when the slow-moving bar was used f or training (Fig. 4b, bottom panel). Figure 4 Asymmetric modification of the tectal receptive field by training with moving stimuli. Full legend High resolution image and legend (73k) Direction-sensitive tectal responses induced by training may result from changes in the tectum or in the retina. We found that when unidirectional fast-bar trai ning was performed under voltage clamp (-80 mV) of the tectal neuron, there was no significant change in the responses of tectal neurons to stimuli in any of th e four directions (Fig. 5a). In three experiments, two training sessions were gi ven sequentially, with the tectal cell in voltage clamp (-80 mV) and then in cur rent clamp. Post-training tests showed no change in the tectal responses after t raining in voltage clamp (-10% 10, s.e.m.), but a significant enhancement after training in current clamp (48% 11%). Thus the induction of direction-sensitive responses required depolarization of the tectal cell. Furthermore, training sti muli that consistently failed to elicit spike trains in the tectal cell did not induce responses with significant directional sensitivity (n = 9, data not shown ). The requirement of postsynaptic depolarization is reminiscent of hebbian LTP observed at many central synapses27, 28, including LTP of these retinotectal syn apses induced by direct electrical stimulation of the retina18 or by dimming lig ht stimuli29. However, step depolarization of the tectal cell (voltage-clamped f rom -70 to 0 mV, 700 ms), when paired with each training stimulus, was not suffi cient to enhance the response to the training stimulus (Fig. 5a). Thus postsynap tic spike-associated depolarization and perhaps other factor(s) are essential fo r the observed training effect. Figure 5 Synaptic mechanisms. Full legend High resolution image and legend (61k) In many regions of the brain27, 28, including the developing tectum18, activity- induced LTP requires activation of NMDA receptors. To test further whether the m echanism underlying changes to the receptive field described here involves NMDA receptor-dependent synaptic plasticity in the tectum, we perfused the tectum loc ally with the NMDA receptor antagonist D-2-amino-5-phosphonopentanoic acid (APV) . The effectiveness of the local perfusion of drugs was shown by the finding tha t perfusion with CNQX (6-cyano-7-nitroquinoxaline-2,3-dione) abolished fast tran sient currents associated with moving-bar-induced CSCs, and that perfusion with APV blocked NMDA receptor-mediated synaptic currents induced by electrical stimu lation of retinal ganglion cells (Fig. 5b, top panel). Perfusion of APV did not affect significantly the tectal response to each sweep of the moving bar, in ter ms of either total CSC charge or spiking activity (Fig. 5b, middle and bottom le ft panels), but it abolished completely direction-specific change in the tectal responses induced by training (Fig. 5b, bottom right panel). Thus, activation of NMDA receptors is required for the induction of direction-sensitive responses, consistent with the blocking effect of APV on LTP of retinotectal synapses18. Further analysis was carried out to determine the relative contributions of exci tatory and inhibitory inputs to the CSCs evoked by each moving-bar stimulus. On the basis of measurements of the average CSC charge at two different clamping vo ltages and the reversal potentials for Na+ and Cl- currents (see Methods), we fo und that excitatory glutamate-mediated inputs contributed to about 90% of the to tal CSC charge at -70 mV (Fig. 5c, left). After training with fast bars, there w as selective increase in these excitatory inputs in terms of the total charge of excitatory components of the CSCs (Fig. 5c, right), as well as a significant in crease in both the size and the average frequency of excitatory postsynaptic cur rents (EPSCs) associated with CSCs evoked by test bars of the training direction (Fig. 5d). Taken together, these results suggest that the training had induced a potentiation of glutamate-mediated inputs on these tectal cells. The observed induction of direction-sensitivity of tectal neurons could result f rom a strengthening of retinotectal connections made by direction-selective reti nal ganglion cells at this early stage of development. Alternatively, it may ari se from a training-induced circuit modification within the tectum. In principle, direction-selectivity of a visual neuron can develop through an activity-depend ent asymmetric circuit modification that results in excitation of the neuron onl y when the stimulus of the preferred direction is presented21-24. We note that, after training with the moving bar, the profiles of CSCs were asymmetrically alt ered for responses elicited by stimuli in the trained and the opposite direction s (Fig. 5e), suggesting distinct circuit modifications for detection of moving s ignals in preferred and non-preferred directions. The observed requirement of te ctal-cell spiking is consistent with a circuit modification that involves spike- time-dependent modification of retinotectal and/or intra-tectal connections, alt hough training-induced changes in the retina may also occur. Although the precis e loci of circuit changes remain to be determined, the rapidity and persistence of receptive field modification shown here illustrate the susceptibility of deve loping neural circuits to the instructive influence by sensory inputs of specifi c spatiotemporal patterns. Methods Tadpole preparation and electrophysiology Xenopus laevis tadpoles of Nieuwkoop a nd Faber stage 42–45 were anaesthetized with saline containing 0.02% MS222 (Sigma) and secured by insect pins to a sylgard-coated dish, and incubated in HEPES-buffered saline containing (in mM): 115 NaCl, 2 KCl, 10 HEPES, 3 CaCl2, 10 glucose, 1.5 MgCl2, and 0.005 glycine (pH 7.4). For recording, the skin was removed and the brain was split open along the midline to expose the inner surface of the tectum. A low dose of -bungarotoxin (2 mg ml-1) was applied to the bath to prevent muscle contraction. As shown previously18, this toxin treatment did not significantly affect the retinotectal responses. The method of perforated-patch, whole-cell recording has been described previously30. The micropipettes were made from borosilicate glass capillaries (Kimax), had a resistance in the range of 3?C4 M, and were tip-filled with internal solution and then back-filled with inter nal solution containing 200 mg ml-1 amphotericin B. The internal solution contai ned (in mM): 110 K-gluconate, 10 KCl, 5 NaCl, 1.5 MgCl2, 20 HEPES, 0.5 EGTA (pH 7.3). Experiments were performed at room temperature (22 °C) and the bath was c onstantly perfused with fresh recording medium at a slow rate (1 ml min-1). Reco rding was made with a patch-clamp amplifier (Axopatch 200A; Axon Instruments). T he whole-cell capacitance was fully compensated and the series resistance (10–20 M) was compensated at 75?C80% (lag 60 μs). Signals were filtered at 5 kHz and sampled at 10 kHz usi ng Axoscope software (Axon Instruments). Local perfusion of the tectum was carri ed out with a glass pipette (opening 30–40 μm) placed near the tectum. All drugs were from Sigma. Concentrations used were: 10 μM CNQX, 10 μM bicuculline, 50 μM D-APV. For mapping of the receptive field, recordings were made under the reversal pote ntial for Cl- current (Ei = -45 to -60 mV), which was determined by the disappea rance and the reversal of spontaneous GABA (-aminobutyric acid)-mediated synapti c currents as the holding potential was changed towards more depolarized values. To determine the contribution of excitatory and inhibitory components of stimul us-evoked CSCs, recordings were repeated at two holding potentials (-90 and -70 mV). The Na+ and Cl- conductances were determined by the formula I(t) = (Ge + ge (t))*(V - Ee) + (Gi + gi(t))*(V - Ei), where I(t) is the amplitude of CSCs at ti me t; ge(t) and gi(t) are Na+ and Cl- conductances; V is the clamping voltage; E e and Ei are reversal potentials for Na+ and Cl- currents, respectively; and Ge and Gi are leakage conductances (which are negligible). Measurements of I(t) at two voltages (averages of 20 repeats) yielded ge(t) and gi(t), and the relative contribution of Na+ and Cl- currents to CSCs were calculated. For estimates of c hanges in EPSCs, fast transients of CSCs were identified by custom-made LabVIEW (National Instruments) software to determine the peak amplitude of individual EP SCs and their frequency. For analysis of spiking patterns, the peak of spikes ob served in the tectal responses was identified and the inter-spike intervals were calculated. Visual stimulation The tectal cell was patched under visual control and the reti na was flattened and stabilized with a glass coverslip after removal of the lens . A small LCD screen (from a virtual reality goggle, Sony, PLM-A35) was mounted on the camera port of the microscope, allowing projection of computer-generated images onto the retina (Fig. 1a). The diameter of the retina was in the range of 250–300 μm and that of the tectal receptive field in the range of 100–2 00 μm. Visual stimulation and analysis software was custom-made. For recep tive field mapping, the entire field for image projection (200 200 μm) wa s divided into a 8 7 grid. White squares (corresponding to each element of the grid) were flashed for 1.5 s in a random sequence, with 5-s intervals. For movin g-bar stimuli, white bars (20 μm in width at the retina) swept across the screen at a speed of 0.3, 0.2 or 0.1 μm ms-1 (fast, medium and slow speed, respectively). For testing tectal responses, bars moving in four orthogonal dir ections (in the sequence of up, right, down and left) were presented with 10-s i ntervals, with 3–5 repeats in each testing session. Two or three testing sessions (with 5-min interval) were performed during the control period. In some experiments, responses to two different speeds were tested. The training session consisted of 60 sweeps of a randomly chosen direction (0.2 Hz), or 240 sweeps of four orthogonal directions with a randomized order (0.5 Hz). For random flashing stimulus, white squares of the same width as the moving bar were flashed at random at non-overlapping locations (see example images in Fig. 3c, inset), exposing the retina to the same total amount of light at any given moment, and covering the same total area as the fast-moving bar stimulus. A total of 120 sets of stimuli (0.2 Hz) were applied for training. For rapid mapping of training-induced changes in the receptive field, the image projection field was divided into a set of ten horizontal and ten vertical bars. Each bar was flashed for 1.5 s (at 5-s interval) and the sequence of flashing bars was randomized. After preliminary mapping of the receptive field, the size of the image projection field was sometimes slightly re-adjusted to cover the entire receptive field. The receptive field was then mapped in two sessions (four repeats in each session) before and after training. Received 6 May 2002;accepted 12 July 2002 References 1. Wiesel, T. N. Postnatal development of the visual cortex and the i nfluence of environment. Nature 299, 583-591 (1982) | PubMed | 2. Katz, L. C. & Shatz, C. J. Synaptic activity and the construction of cortical circuits. Science 274, 1133-1138 (1996) | Article | PubMed | 3. Penn, A. A. & Shatz, C. J. Brain waves and brain wiring: the role of endogeno us and sensory-driven neural activity in development. Pediatr. Res. 45, 447-458 (1999) | PubMed | 4. Zhang, L. I. & Poo, M. M. Electrical activity and development of neural circu its. Nature Neurosci. 4 Suppl., 1207-1214 (2001) | Article | PubMed | 5. Hubel, D. H. & Wiesel, T. N. Binocular interaction in striate cortex of kitte ns reared with artificial squint. J. Neurophysiol. 28, 1041-1059 (1965) | PubMed | 6. Meister, M., Wong, R. O., Baylor, D. A. & Shatz, C. J. Synchronous bursts of action potentials in ganglion cells of the developing mammalian retina. Science 252, 939-943 (1991) | PubMed | 7. Penn, A. A., Riquelme, P. A., Feller, M. B. & Shatz, C. J. Competition in ret inogeniculate patterning driven by spontaneous activity. Science 279, 2108-2112 (1998) | Article | PubMed | 8. Cynader, M., Berman, N. & Hein, A. Cats reared in stroboscopic Illumination: effects on receptive fields in visual cortex. Proc. Natl Acad. Sci. USA 70, 1353 -1354 (1973) | PubMed | 9. Schmidt, J. T. & Eisele, L. E. Stroboscopic illumination and dark rearing blo ck the sharpening of the regenerated retinotectal map in goldfish. Neuroscience 14, 535-546 (1985) | PubMed | 10. Weliky, M. & Katz, L. C. Disruption of orientation tuning in visual cortex b y artificially correlated neuronal activity. Nature 386, 680-685 (1997) | PubMed | 11. Sharma, J., Angelucci, A. & Sur, M. Induction of visual orientation modules in auditory cortex. Nature 404, 841-847 (2000) | Article | PubMed | 12. von Melchner, L., Pallas, S. L. & Sur, M. Visual behaviour mediated by retin al projections directed to the auditory pathway. Nature 404, 871-876 (2000) | Ar ticle | PubMed | 13. Fregnac, Y., Shulz, D., Thorpe, S. & Bienenstock, E. A cellular analogue of visual cortical plasticity. Nature 333, 367-370 (1988) | PubMed | 14. Schuett, S., Bonhoeffer, T. & Hubener, M. Pairing-induced changes of orienta tion maps in cat visual cortex. Neuron 32, 325-337 (2001) | PubMed | 15. Gaze, R. M., Keating, M. J. & Chung, S. H. The evolution of the retinotectal map during development in Xenopus. Proc. R. Soc. London B 185, 301-330 (1974) 16. Holt, C. E. & Harris, W. A. Order in the initial retinotectal map in Xenopus : a new technique for labelling growing nerve fibres. Nature 301, 150-152 (1983) | PubMed | 17. Markram, H., Lübke, J., Frotscher, M. & Sakmann, B. Regulation of synaptic efficacy by coincidence of postsynaptic APs and EPSPs. Science 275, 213-215 (199 7) | Article | PubMed | 18. Zhang, L. I., Tao, H. W., Holt, C. E., Harris, W. A. & Poo, M. M. A critical window for cooperation and competition among developing retinotectal synapses. Nature 395, 37-44 (1998) | Article | PubMed | 19. Feldman, D. E. Timing-based LTP and LTD at vertical inputs to layer II/III p yramidal cells in rat barrel cortex. Neuron 27, 45-56 (2000) | PubMed | 20. Boettiger, C. A. & Doupe, A. J. Developmentally restricted synaptic plastici ty in a songbird nucleus required for song learning. Neuron 31, 809-818 (2001) | PubMed | 21. Abbott, L. F. & Blum, K. I. Functional significance of long-term potentiatio n for sequence learning and prediction. Cereb. Cortex 6, 406-416 (1996) | PubMed | 22. Zanker, J. M. & Zeil, J. Motion Vision--Computational, Neural, and Ecologica l Constraints (Springer, New York, 2000) 23. Rao, R. P. & Sejnowski, T. J. Predictive learning of temporal sequences in r ecurrent neocortical circuits. Novartis Found. Symp. 239, 208-229 (2001) | PubMe d | 24. Roberts, P. D. Computational consequences of temporally asymmetric learning rules. I. Differential hebbian learning. J. Comput. Neurosci. 7, 235-246 (1999) | Article | PubMed | 25. Mehta, M. R., Barnes, C. A. & McNaughton, B. L. Experience-dependent, asymme tric expansion of hippocampal place fields. Proc. Natl Acad. Sci. USA 94, 8918-8 921 (1997) | Article | PubMed | 26. Mehta, M. R., Quirk, M. C. & Wilson, M. A. Experience-dependent asymmetric s hape of hippocampal receptive fields. Neuron 25, 707-715 (2000) | PubMed | 27. Bliss, T. V. P. & Collingridge, G. L. A synaptic model of memory: long-term potentiation in the hippocampus. Nature 361, 31-39 (1993) | PubMed | 28. Malenka, R. C. & Nicoll, R. A. Long-term potentiation--a decade of progress? Science 285, 1870-1874 (1999) | Article | PubMed | 29. Zhang, L. I., Tao, H. & Poo, M. Visual input induces long-term potentiation of developing retinotectal synapses. Nature Neurosci. 3, 708-715 (2000) | Articl e | PubMed | 30. Rae, J., Cooper, K., Gates, P. & Watsky, M. Low access resistance perforated patch recordings using amphotericin B. J. Neurosci. Methods 37, 15-26 (1991) | PubMed | Acknowledgements. This work was supported by grants from NS