Survey
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
Springer-Verlag 1998 Exp Brain Res (1998) 118:489±500 RESEARCH ARTICLE Lionel G. Nowak ´ Jean Bullier Axons, but not cell bodies, are activated by electrical stimulation in cortical gray matter II. Evidence from selective inactivation of cell bodies and axon initial segments Abstract The results presented in the companion paper showed that extracellular electrical stimulation of the gray matter directly activates axons, but not cell bodies. The second set of experiments presented here was designed to separate the contribution of the axon initial segments and cell bodies from that of the axonal branches to the pool of presynaptic neuronal elements activated by electrical stimulation. For that purpose, N-methyl-d-aspartate (NMDA) iontophoresis was used to induce a selective inactivation of the cell body and of the adjoining portion of the axon by depolarization block, without affecting axonal branches that lack NMDA receptors. After NMDA iontophoresis, the neurons located near the iontophoresis electrode became unable to generate action potentials in an irreversible manner. When the NMDA-induced depolarization block was performed at the site of electrical stimulation, an unexpected increase in the amplitude of the orthodromic responses was observed. Several control experiments suggested that the field potential increase was due to changes of the local environment in the vicinity of the iontophoresis pipette, which led to an increased excitability of the axons. After the period of superexcitability, the orthodromic responses displayed an amplitude that was 15Ð20% lower than that observed before the NMDA-induced depolarization block, even though cell bodies and axon initial segment at the site of stimulation could not be activated by electrical stimulation. This result shows a low contribution for axon initial segments to the pool of neuronal elements activated by the electrical stimulation. Altogether, these experiments demonstrate that the postsynaptic responses obtained after electrical stimulation of the cortical gray matter result almost exclusively from the activation of axonal branches. Since the neocortex is organised as a network of local and long- ) L.G. Nowak1 ´ J. Bullier ( ) INSERM UnitØ 371, ªCerveau et Vision,º 18 Avenue du Doyen LØpine, F-69500 Bron, France Present address: Section of Neurobiology, Yale University School of Medicine, C303 Sterling Hall of Medicine, 333 Cedar Street, New Haven, CT 06510, USA 1 range reciprocal connections, great attention must be paid to the interpretation of data obtained with electrical stimualtion. Key words Visual cortex ´ Brain slice ´ Intracortical microstimulation ´ NMDA ´ Rat Introduction The results reported in the companion paper (Nowak and Bullier 1997) suggest that, when a post synaptic response is observed after an extracellular electrical stimulation, axons, rather than cell bodies, are the presynaptic elements activated by extracellular electrical stimulation. This, we argued, results from the different biophysical properties of axons and cell bodies, endowing axons with a lower threshold for activation. However, the relative contribution of axon initial segments and that of axonal branches could not be determined from the chronaxie measurements used in that study. The aim of the experiment reported in the present paper was to determine in what proportion axon initial segments and axonal branches are activated by extracellular electrical stimulation. For that purpose, we chose to compare the effects of electrical stimulation before and after the inactivation of one of these two groups. To achieve this comparison, we used a method that prevents spike initiation at the level of the cell body and initial segment, while leaving undisturbed the generation and propagation of action potentials in axonal branches. This method consisted in applying NMDA by iontophoresis at the site of stimulation. It may look somewhat paradoxical to use NMDA to inactivate neurons, but the following explanation gives the rationale for such a choice. Dendrites and possibly cell bodies possess glutamate receptors that are activated by NMDA (reviewed in Huntley et al. 1994b). The reversal potential for NMDA-gated channels is about 0 mV (Crunelli et al. 1984; Nowak et al. 1984). On the other hand, sodium channels of neocortical neurons are inactivated at a potential of 40 mV (Huguenard et al. 1988). There- 490 fore, NMDA can depolarize neurons by such an amount that sodium channels are inactivated. In that situation, dendrites and cell bodies are in a state of ªdepolarisation block.º It follows that extracellularly applied electrical stimulation is unable to elicit any action potential. The initial segment should also be blocked, and presumably also the first node of Ranvier, because they are electrotonically close to the cell body (Gogan et al. 1983). On the contrary, axonal branches, which do not possess glutamate receptors, should not be affected. If electrical stimulation is applied during NMDA-induced depolarization block, only axonal branches should be activated. Then, by comparing results obtained with electrical stimulation applied before, during, and after the NMDA-induced depolarization block, it was possible to determine whether initial segments on the one hand, or axonal branches on the other hand, are the elements activated by electrical stimulation. The conclusion we finally reached is that the postsynaptic effects obtained after electrical stimulation in the cortical gray matter result almost exclusively from the activation of axonal branches. Materials and methods Brain slice preparation The procedure for the preparation of cortical brain slices is the same as the one described in the preceding paper. Recording and stimulation Recording and stimulation were performed in areas 17 and 18a of rat visual cortex. Some of the results presented in this paper were obtained from neurons involved in corticocortical connections between these two areas, identified as described in Nowak et al. (1997). Extracellular recording of field potentials (Figs. 3Ð5) was made with glass micropipettes filled with 1 M NaCl (d.c. resistance 1Ð5 MW). Field potentials were amplified on an amplifier (Biologic VF 180) followed by a Neurolog device containing amplifiers and filters. The field potentials were low- and high-pass filtered between 0.1 or 1 Hz and 1 khz. The extracellular recordings of the action potentials and field potentials presented in Figs. 1 and 2 were obtained with tungsten-in-glass microelectrodes (Merrill and Ainsworth 1972) with 15 to 25-m exposed tips and plated with platinum black (impedance less than 0.5 MW at 1000 Hz).Amplification and filtering were achieved through a Neurolog recording system. Filtering was adjusted for individual recordings to give a proper signal-to-noise ratio. Tungsten-in-glass microelectrodes plated with platinum black were also used for electrical stimulation. The glass was removed over a length of 15Ð25 m. Electrical stimulation consisted of monopolar, cathodal pulses of 0.2 ms duration. Single pulses were delivered at a frequency of 0.5 Hz or 0.3 Hz through a stimulation isolation unit (Neurolog). Iontophoresis NMDA (Sigma) was dissolved in distilled water to a final concentration of 200 mM (pH adjusted to 8.0 with NaOH). Iontophoresis was made through double-barreled glass micropipettes. One barrel was filled with NMDA and the other, used for return current, with 1 M NaCl. Tungsten-in-glass microelectrodes, used for stimulation and/or recording, were glued to iontophoresis micropipettes with dental ce- ment. The distance between the tips varied between 20 and 150 m. For small tip separations, we noticed that the leakage of NMDA could induce action potentials and eventually a depolarization block. These effects were avoided by using a retaining current of 20 nA. Results Distance at which NMDA induces a depolarisation block Before starting the experiments themselves, two sets of control experiments had to be done. Since NMDA was iontophoretically applied, the first series of control experiments was designed to determine the radius of the sphere within which neurons are in depolarization block for a given current of iontophoresis. The set-up for such experiments is shown in Fig. 1A. Action potentials were recorded through tungsten-in-glass microelectrodes that were glued to the iontophoresis micropipettes. The lateral separation between the microelectrode and the pipette tips was 60, 120, and 150 m for three electrode assemblies from which the results shown in Fig. 1 were obtained. For the recordings obtained with the electrode assembly of 150 m tip separation, electrical stimulation was applied in the white matter or in the lower cortical layers to induce orthodromic or antidromic action potentials. For the other two electrode assemblies, action potentials were produced by the leakage of NMDA or by applying small amounts of ejection current. Once action potentials were elicited by one of these two modes, the iontophoresis current was raised in steps of 5 nA or 10 nA until action potentials were suppressed. One example is shown in Fig. 1B. Each trace corresponds to the superposition of 20 sweeps. Electrical stimulation was applied in the white matter and elicited orthodromic action potentials. The large action potentials disappeared with a iontophoresis current of 30 nA. They recovered after the iontophoresis apparatus was switched back to retaining mode. Another example is illustrated in Fig. 1C. An antidromic action potential was produced by white matter stimulation. The antidromic action potential was suppressed by a iontophoresis current of 40 nA. The recovery trace shows that the action potentials have a smaller amplitude and a slightly longer latency than in the control situation. This is not unexpected, since both latency and amplitude of antidromic action potentials can be modified by changes of the neuron input resistance and resting potential (see Lipski 1981), both effects that NMDA is likely to produce before its complete washout. Figure 1B and C shows in addition that, during the NMDA-induced depolarization block, the action potential is entirely suppressed, not simply reduced in amplitude. This indicates that the axon initial segment also lost its ability to generate action potentials. Figure 1D presents the distribution histogram of the iontophoresis current leading to a depolarization block, measured with the electrode assemblies having separations of 120 m and 150 m. The pale gray bars corre- 491 Fig. 1A±D Distance at which NMDA produces a depolarization block. A Set-up. A tungsten-in-glass microelectrode for extracellular recording and a micropipette for NMDA iontophoresis were glued together, such that the separation between their tips was perfectly controlled. Recording and iontophoresis were limited to the supragranular layers. The axis defined by the line passing at the tip of both electrodes was parallel to the pial surface, so that recording and iontophoresis were performed in the same cortical layer. B One example of NMDA-induced depolarization block. The separation between the tips of the recording microelectrode and the iontophoresis pipette was 150 m. Orthodromic action potentials were elicited by white matter stimulation (upper trace). NMDA applied with an iontophoresis current of 30 nA led to their suppression (middle trace). They recovered (lower trace); however, they are less grouped than during the control period. Although they were not tak- spond to the electrode assembly for which the tip separation was 120 m. In that case, the neurons were activated by the NMDA itself. The dark gray and the hatched bars correspond to the electrode assembly for which the tip separation was 150 m. The action potentials in that case were induced by electrical stimulation. These were either en into account owing to uncertainty relative to their separation from both iontophoresis and recording electrodes, one can observe that the small long-latency action potentials visible in the control period did not recover. Another example of NMDA-induced depolarization block appears in C. In that case the white matter stimulation produced an antidromic action potential. About ten individuals sweeps were averaged for the different periods and are shown superimposed. The antidromic action potential was suppressed with a iontophoresis current of 40 nA. D Histogram illustrating the number of neurons in which NMDA induced a depolarization block as a function of the iontophoresis current required to produce this effect. Data presented are from two electrode assemblies having 150 m or 120 m separation between the recording and the iontophoresis tips. In this and the following figures, positive potential is upward antidromic action potentials (hatched bars) or orthodromic action potentials (dark gray bars). The histogram shows that one-half of the cells were in a state of depolarization block with iontophoresis current equal to 30 nA when the separation was 120 m, and 40 nA when the separation was 150 m. It further shows that 492 Fig. 2A, B Time course of the NMDA-induced depolarization block. A Example of an antidromic population spike and its suppression by NMDA. Each trace corresponds to the mean of ten sweeps. One trace was obtained every 5 min. Time is indicated on the right of the series of field potentials every two traces. The control period was established between 0 and 30 min. NMDA was iontophorized between 30 and 40 min with an ejection current of 80 nA. The NMDA led to an almost complete suppression of the antidromic population spike. The recovery period corresponds to time 40 to 110 min. The antidromic population spike did not recover. B Line chart summarizing the effect of NMDA on antidromic population spikes in seven experiments. The control period was always 30 min. The period of NMDA application (30±40 min) is indicated by the two vertical dashed lines. The recovery period was at least 30 min (40Ð70 min) but was longer for two experiments. The bars show 1 SEM. The horizontal dashed line corresponds to the mean control value (100%). NMDA led in all cases to a strong reduction of the population spike amplitude, without recovery all neurons located at a distance of 150 m from the iontophoresis electrode were inactivated with a current of 80 nA. With the electrode assembly having a tip separation of 150 m, the action potentials that were antidromically elicited were suppressed with an iontophoresis current similar to that required for the suppression of orthodromic action potentials (Mann-Whitney U-test, P=0.9). With the electrode assembly having a tip separation of 60 m (not shown in the histogram), all the neurons became unable to generate action potentials with a iontophoresis current of 20 nA only. Time course of the NMDA-induced depolarization block The second control experiment consisted in determining the duration of the NMDA-induced depolarization block. The set-up used for these experiments was the same as the one presented in Fig. 1. The only difference was that the filtering of the recording obtained through the tungstenin-glass microelectrode was modified to allow the measurement of antidromic population spikes (the simulta- neous antidromic activation of a large number of cells) produced by electrical stimulation of the white matter of the lower cortical layers. One example appears in Fig. 2A. A control baseline was first established, corresponding to the antidromic population spikes obtained from time 0 to 30 min. After the last recording of the control period (30 min), the iontophoresis was switched from its retaining mode to the ejection mode. The ejection current was 80 nA and was maintained for 10 min. During the iontophoresis of NMDA, the antidromic population spikes were almost completely suppressed (traces at 35 min and 40 min). After the 10 min of NMDA application, the ionotophoresis was switched back to its remaining mode. Recordings obtained at time 45±110 min in Fig. 2A show that there was no recovery of the initial population spikes. We determined the time course of the NMDA-induced depolarization block in seven cases. The distance between the tip of the iontophoresis micropipette and that of the recording microelectrode was 24 m in one case, 35 m in three cases, 60 m in two cases, and 120 m in one case. Figure 2B illustrates the pooling of these seven ex- 493 periments. The maximum amplitude of the antidromic population spikes was first measured for each individual case. The mean of the control values (obtained between 0 and 30 min) was thereafter calculated and used to normalize the amplitudes. Once expressed as percentages, the mean of the seven amplitudes obtained at a given time was calculated. Figure 2B shows that the iontophoresis of NMDA led to a very strong decrease in the amplitude of antidromically elicited population spikes in all the cases. This decrease persisted during the whole period following iontophoresis. For five of the seven cases, this period lasted 30 min (40Ð70 min in Fig. 2B). For another case, it lasted 60 min, and for the seventh, which is illustrated in Fig. 2A, it lasted 70 min. Field potentials were not completely suppressed, however. There remained about less than 10% of the control amplitude. This remaining activity could represent a contribution of action potentials recorded from axons. Alternatively, it could correspond to the contribution of action potentials from neurons that are far enough to be unaffected by NMDA. If this was the case, the neurons generating these action potentials must have been located at a distance larger than 150 m from the iontophoresis micropipette, since, as shown in Fig. 1, an iontophoresis current of 80 nA suppressed the single-unit activities up to that distance at least. The lack or recovery is at variance from the examples shown in Fig. 1B and C. However, the recovery of singleunit action potentials was observed only if the iontophoresis current was applied for a short period corresponding to the time necessary to check that the action potentials were suppressed. If the iontophoresis was extended to a few minutes, there was no recovery of single-unit activities. Therefore, when iontophorized for 10 min with 80 nA, the action of NMDA was not simply a transient inactivation, since neurons definitively lost their ability to generate action potentials. Rather, NMDA induced a microlesion of the neurons, which proved to be irreversible. Stimulation during NMDA-induced depolarization block The set-up of these experiments is shown in Fig. 3A. The electrical stimulation was applied at the site of NMDA iontophoresis (labeled ªstim.º in Fig. 3A). In all the cases stimulation and iontophoresis were made in the supragranular layers. The separation between the tips of the stimulating electrode and iontophoresis pipette was between 20 and 35 m. For this series of experiments (n=12), electrical stimulation intensity was 15 A in nine cases and 12 A in the three other cases. The electrical stimulation must activate neuronal elements contained in an area smaller than the area in which NMDA induces a depolarization block. According to experiments reported in another paper (Nowak and Bullier 1996), a stimulating intensity of 15 A should activate neuronal elements contained within a sphere having a maximum radius of 82 m. The dark gray area surround- ing the stimulating and iontophoresis electrodes in Fig. 3A schematizes this area. The light gray area in Fig. 3A schematizes the area in which neurons are in a state of depolarization block with 80 nA of iontophoresis current. According to the experiments illustrated in Fig. 1, this area has a minimum radius of 150 m. Electrical stimulation was used to elicit orthodromic field potentials. All recordings were made in the supragranular layers. The sets of connections that were activated by electrical stimulation in these experiments were either the intrinsic horizontal connections (n=6, separation between recording and stimulating sites ranging from 0.5 to 0.9 mm) or the corticocortical connections (n=6, separation between recording and stimulating sites ranging from 1.1 to 1.5 mm). If electrical stimulation effectively activates axon initial segments, cell bodies, and dendrites, then the amplitude of the orthodromic field potential should be reduced when the stimulation is applied during the NMDA-induced depolarization block. Unexpectedly, the opposite was observed. One series of orthodromically elicited field potentials is presented in Fig. 3B. After a control period of 30 min, NMDA (80 nA) was applied for 10 min. There was not a decrease, but an increase in the field potential amplitude during the NMDA induced depolarization block (time 35 min and 40 min). In other words, when neurons (dendrites, cell bodies, and initial segments) are in such a state that they cannot be activated by the electrical stimulation, the amplitude of the field potential is larger than when they could contribute to the pool of neuronal elements activated. During the following 30 min of recovery, the amplitude of the field potential declined and reached its initial control value at time 60 min. The data pooled from the 12 experiments are shown in Fig. 3C. The maximum amplitude of the field potential was measured and normalized as already described. Due to the small stimulation intensity and the large separation between stimulating electrode and recording micropipette, the field potential amplitude was often small and averaged 160109 (SD) V during the control period. In all the cases, the amplitude increased during the NMDA iontophoresis, by a factor larger than 1.5 on average. The increase in amplitude was highly significant. The amplitude of the field potentials returned to its control value 20 min after the end of the iontophoresis. The amplitudes measured at 60, 65 and 70 min are not significantly different from those of the control period (unpaired t-test, P>0.05). For 5 of the 12 cases, however, the recovery period was prolonged from 30 to 50 min. This extended recovery period is illustrated by the gray dots in Fig. 3C. It can be seen that, when the amplitude reaches a stable value, it is significantly smaller than the control value (unpaired t-test, P<0.05 for time 75Ð90 min). These experiments show that, when there was hardly any cell body or initial segment that could be activated at the site of stimulation (Fig. 2), the amplitude of the orthodromic field potential was decreased by 15Ð20% only compared with what it was before the NMDA induced microlesion. This suggests that the contribution of dendrites, cell bodies, and axon initial segments to the pool of presynaptic neuronal el- 494 Fig. 3A±D Stimulation during NMDA-induced depolarization block. The set-up is illustrated in A. The stimulation (Stim) was applied at the site of NMDA iontophoresis. The dark gray circle near the tip of the stimulating-iontophoresis electrode assembly schematizes the sphere in which neuronal elements are activated with a current of 15 A or less, which is smaller than the sphere in which neurons are led to a depolarization block with a iontophoresis current of 80 nA (schematized by the light gray area). An example of the changes of an orthodromic field potential when stimulation is applied at the site of NMDA ejection before, during and after the iontophoresis appears in B. The recordings obtained between 0 and 30 min correspond to the control period and were used to establish the baseline value. Iontophoresis of NMDA (80 nA) was performed between 30 and 40 min. Time 40±70 min corresponds to the recovery period. C Line chart summarizing the evolution of the orthodromic field potential amplitude measured every 5 min in 12 experiments. The dots represent the mean amplitude expressed as percentage of the baseline value. The bars indicate SEM. The gray dots correspond to five experiments for which the recovery period was prolonged from 30 to 50 min. D Evolution of the peak latency of the orthodromic field potentials. It appears unaffected by the NMDA iontophoresis ements activated by electrical stimulation was much lower than the contribution of axonal branches, which accounts for 80Ð85% of the postsynaptic response observed. Origin of the transient increase in amplitude Because the increase in amplitude of field potentials observed during the iontophoresis of NMDA was unexpected, we designed additional experiments to determine its origin. We first reasoned that the NMDA, spreading from the iontophoresis pipette, could reach neurons located outside the region of depolarization block and depolarize them in a tonic fashion. This depolarization could facilitate their orthodromic activation by the axons stimulated near the stimulating electrode that project onto them. If, in addition, these neurons were sending projections to the neurons we were recording, their activation would lead to the activation of additional synapses, thereby to an increase in field potential amplitude. Nevertheless, this 495 Fig. 4A±C Evolution of orthodromic field potentials when electrical stimulation is applied at a site remote from the site of NMDA-induced depolarization block. Set-up illustrated in A. The stimulation (Stim.) was applied at the site remote from the site of NMDA iontophoresis. B Time series of a field potential obtained when the electrical stimulation was applied in the infragranular layers. The pooled data from five experiments are presented in C. In all five cases the stimulation was applied in the infragranular layer in an interleaved fashion with those applied at the site of NMDA iontophoresis (5 of the 12 cases reported in Fig. 3C). The dots in C represent the mean of the amplitude after normalization with respect to the baseline value. The bars indicate 1 SEM. The increase in field potential amplitude was not observed when stimulation was applied at a site remote from the site of NMDA iontophoresis increased amplitude would result from the recruitment of a disynaptic pathway, implying that the latency of the maximum amplitude of the field potential should increase. Therefore, to determine whether NMDA led to the recruitment of a disynaptic pathway, we measured the latency of the maximal amplitude of the field potentials (peak latency). The case illustrated in Fig. 3B does not show the expected change in latency. The pooled data for the 12 experiments is illustrated in Fig. 3D. The mean latency remains very close to 100%, during both the NMDA iontophoresis and the recovery period. The peak latency during and after the NMDA iontophoresis was, indeed, not significantly different from its control value (unpaired t-test, P>0.05). This suggests that the transient increase in field potential amplitude was not the consequence of the recruitment of a polysynaptic pathway. Two other possibilities were tested in the same experiments. These two possibilities have in common some properties of cortical neurons, which are their inward retification properties. These properties are such that, when a neuron is depolarized from rest, the amplitude of the EPSP can increase, contrary to what would be predicted if the amplitude of the EPSPs was solely determined by the driving force. These rectifying properties are related to the presence of some voltage-dependent conductances (Stafstrom et al. 1985; Deisz et al. 1991) and to the behavior of glutamate receptors of the NMDA subtype (MacDonald et al. 1982; Flatman et al. 1983; Nowak et al. 1984; Thomson 1986a, b). For these rectifying properties to be turned on, it is required that the neurons we were recording from underwent a tonic subthreshold depolarization during the ionto- 496 Fig. 5A±C Evolution of the amplitude of antidromic population spikes evoked by electrical stimulation applied before, during, and after NMDA-induced depolarization block. A Set-up. The stimulating-iontophoresis electrode assembly was placed in the infragranular layers. The stimulation was used to elicit antidromic population spikes that were recorded in the supragranular layers. An example appears in B. Time 0±30 min corresponds to the control period. NMDA was iontophorized (80 nA) between 30 and 40 min. C Line chart summarizing the time course of the antidromic population spike amplitude before, during, and after the iontophoresis of NMDA. Six experiments are pooled. The dots correspond to the mean amplitude expressed as percentage of baseline value and the bars to 1 SEM. This shows that antidromic population spikes resulting from axonal activation were increased in a way similar to orthodromic field potentials by the iontophoresis of NMDA at the stimulating site phoresis of NMDA. Such tonic subthreshold depolarization could have occurred if the NMDA was spreading from the site of iontophoresis to the recording site, although such a spread may be considered unlikely given the large separation between recording and iontophoresis sites. Alternatively, NMDA could have activated neurons that were in the periphery of the area of depolarization block, which would induce a tonic firing at their level. If these neurons were projecting to the neurons we were recording from, their tonic firing could have induced a subthreshold depolarization at the recording site. In both cases, the depolarization could have triggered the mechanisms responsible for the anomalous rectification. This hypothesis implies that the changes were occurring at the level of the neurons we were recording from Therefore, the amplitude of field potentials should have increased also if the electrical stimulation was applied at a position remote from the iontophoresis micropipette. This hypothesis was tested and the results are presented in Fig. 4. The stimulating electrode was in a site distant from that of iontophoresis, in the lower cortical layers below the recording site (Fig. 4A). A series of field poten- tials obtained before, during, and after the NMDA iontophoresis is shown in Fig. 4B. These field potentials were obtained in an interleaved fashion with those shown in Fig. 3B (This explains why the timing is deplayed by 2.5 min with respect to the timing of the measurements of Fig. 3B). Contrary to those of Fig. 3B, however, the amplitude of the field potentials of Fig. 4B did not increase during the iontophoresis of NMDA, even though the same neurons were recorded. Figure 4C summarizes the results obtained in five similar experiments. In no case did the field potentials amplitude increase during the NMDA iontophoresis. This excludes the subthreshold depolarization and the associated activation of inward rectification in the neurons located at the site of recording as the mechanism responsible for the increase in field potential amplitude that was observed when stimulation and iontophoresis were given at the same site. Neither the spread of NMDA to the site of recording nor the depolarization of neurons at the recording site by a tonic drive from the neurons located at some distance from the iontophoresis electrode are responsible for the transient increase. 497 Therefore, the origin of the transient depolarization must be located at the site of the iontophoresis itself. The massive depolarization caused by NMDA must have led to a modification of the local environment. In particular, there should have been an increase in the extracellular concentration of potassium. It is known that increasing the extracellular concentration of potassium increases the excitability of neurons (Hablitz and Lundervold 1981; Balestrino et al. 1986). It is also known that excitatoxic lesions are accompanied by a swelling of the neuronal processes (Garthwaite et al. 1986; Choi 1987). This swelling might have increased the extracellular impedance, leading to an increased extracellular potential for a current of the same intensity. In both cases, applying NMDA would, indirectly, have acted as if the threshold for activation of neuronal elements was decreased. Figure 5 shows that, in fact, the excitability of axons was increased by the application of NMDA. In these experiments, the set-up of which appears in Fig. 5A, the electrical stimulation and the NMDA were applied in the lower cortical layers while the recordings were obtained in the supragranular layers. Unlike in the preceding cases, we recorded antidromically elicited population spikes that resulted from the activation of axons passing near the stimulating electrode (note the different shape and time course of the responses illustrated in Fig. 5B compared with Figs. 3B, 4B). An example of a time series of population spikes is presented in Fig. 5B. After the control period of 30 min, NMDA was iontophorized (80 nA) for 10 min. Similarly to what was observed with the orthodromic field potentials of Fig. 3B, the application of NMDA also led to an increase in amplitude for the antidromic population spikes. Summary data after pooling of six experiments is given in Fig. 5C. In all the cases, the amplitude of the antidromic population spikes increased when electrical stimulation was applied during the iontophoresis of NMDA. After the end of the iontophoresis, the amplitude decreased toward the control value. However, the control value was not reached after 70 min. This is due to two cases, which retained an amplitude higher at 70 min than during the control period. In one of these two cases, the recovery was prolonged and it was found that the amplitude remained stable at this higher value. In summary, these different experiments suggest that the local environment surrounding the iontophoresis micropipette was modified by the NMDA application. The most likely explanations for the transient increase in both orthodromic and antidromic field potentials are the following: NMDA application might have induced a massive depolarization of the neurons. This would have led to an increase in the extracellular concentration of potassium. This extracellular potassium would have increased the excitability of neuronal elements present in the neighborhood. Alternatively, the application of NMDA might have led to a swelling of the neuronal processes that would have resulted in an increase in the extracellular impedance, hence to a decreased threshold for axonal activation. The consequence of that increased excitability was an in- crease in the amplitude of the responses. After the iontophoresis, as the NMDA was washed out, the amplitude of the responses returned to its baseline value as the concentration of potassium was equilibrated with that of the bathing medium or as the neuronal swelling was resumed. Discussion The experiments based on chronaxie measurements suggested that, when electrical stimulation is applied close to a cell body, its axon initial segment is preferentially activated and problably also a number of axonal branches located in the vicinity of the electrode (Nowak and Bullier 1997). However, the chronaxie measurement did not allow the differentiation between the activation of axonal branches and that of axon initial segments. It could be argued that initial segments are more excitable than axonal processes and may contribute for a large part to the postsynaptic responses elicited by extracellular electrical stimulation. The experiments based on stimulation during NMDA-induced depolarization block, designed to determine this contribution, show that the postsynaptic responses obtained after stimulation of the cortical gray matter results not from the activation of cell bodies or initial segments but almost exclusively from the activation of axonal branches. NMDA suppresses action potential electrogenesis The first requirement for these experiments was that axonal branches are not affected by NMDA. Several immunohistochemical studies have shown that axonal trunks lack NMDA receptors (Aoki et al. 1994; Huntley et al. 1994a; Petralia et al. 1994). That axons are not sensitive to glutamate has also been shown with electrophysiological methods (Christian and Dubek 1988). This is substantiated by the results of Fig. 5, showing no attenuation of antidromic population spike after their transient increase, which also indicate that the current used for iontophoresis did not lead to any axonal damage. The second requirement for these experiments was that NMDA induced a depolarization block in all neurons. Electrophysiological studies have shown that all neocortical pyramidal cells bear NMDA receptors: all respond to NMDA applied iontophoretically and all display components in their EPSPs that are mediated by NMDA receptors (Thomson 1986a, b; Sutor and Hablitz 1989a, b; Thomson et al. 1989; Artola and Singer 1990; Armstrong-James et al. 1993). These electrophysiological evidences are paralleled by several recent labeling and immunohistochemical studies (Benke et al. 1993; Aoki et al. 1994; Huntley et al. 1994a; Petralia et al. 1994) that indicate that the very large majority of cortical neurons possess NMDA receptors. In fact, the only type of neuron that may lack NMDA receptors corresponds to a subpopulation of inhibitory neurons (Huntley et al. 1994a; see also Thomson et al. 1993; Thomson and Deuchars 498 1994). That some inhibitory neurons lack NMDA receptors was not a problem in our case, since we were interested in inactivating excitatory neurons having long-range projections. All the neurons became unable to elicit action potentials after the iontophoresis of a sufficient amount of NMDA (Fig. 1). Our conclusion that NMDA leads to a depolarization block was based on extracellular recordings. A more precise measure of the sodium channel inactivation induced by NMDA would have been obtained using intracellular recording. However, several studies based on intracellular recording have already shown that small amounts of glutamate or NMDA effectively depolarize cells to well above the threshold for action potential initation (Schwarztkroin 1975; Hablitz 1982; Sutor and Hablitz 1989b). The third requirement was that the initial segment was blocked. We also showed that action potentials were not only partially, but completely, suppressed by NMDA. This indicates that the depolarization block induced in the cell body spread to the initial segment that become unable to elicit action potentials. It has been suggested that the action potential initiation site is not necessarily the initial segment, but the first node of Ranvier on the axon (Gogan et al. 1983; Golbert and Johnston 1996). In our experimental situation, it has not been possible to test whether the NMDA-induced depolarization block extended to the first node of Ranvier, although this could have occurred given the long space constant of the axon (Gogan et al. 1983). If this occurred, however, it would have led to an underestimate of the proportion of axons activated by electrical stimulation, which is already of 80Ð85%. One of the reasons why we chose NMDA is that, being an artificial agonist, its spread is not limited by uptake systems. We indeed observed that small amounts of NMDA were able to suppress action potential generation at a relatively large distance from the iontophoresis micropipette (Fig. 1). There was no recovery of action potential electrogenesis after 10 min of NMDA iontophoresis (Fig. 2) even after 1 h of recovery. That excessive glutamate or NMDA concentration is neurotoxic has been demonstrated in a number of studies (Rothman 1985; Garthwaite et al. 1986; Choi 1987; Lehmann and Jacobson 1990; McNamara and Dingledine 1990). The effects produced by NMDA corresponded to a chemical microlesion in a restricted part of the cortex. Stimulation during and after NMDA-induced cortical microlesion When NMDA was used to suppress action potential generation in cell bodies and axon initial segments at the site of stimulation, there appeared to be an unexpected increase in the orthodromic field potential amplitude. The control experiments (Figs. 4, 5) showed that the increase was due to changes at the level of the iontophoresis electrode. The most probable change that occurred was a modification of the local concentration of ions or a change in the local impedance of the extracellular medi- um, which led to an increased excitability for the neuronal elements present in the vicinity. The transient increase in the field potentials observed during the NMDA iontophoresis obscured the effect of the inactivation. Another reason why we chose NMDA is the irreversibility of its effects. After the period of superexcitability, the stimulation yielded field potentials that were lower by 15Ð20% only than those obtained during the control period (Fig. 3). This strongly suggests that the activation of axonal branches is responsible for the bulk (around 80Ð85%) of the orthodromic responses observed after electrical stimulation in the cortical gray matter. Our conclusions are at variance from those obtained by Hirsch (1995). To determine the contribution of axonal branches to the pool of neuronal elements activated by electrical stimulation, she used local application of GABA near the stimulating electrode to raise the activation threshold for cell bodies. In her Fig. 1 (upper part), the GABA application led to the disappearance of evoked synaptic responses. Nevertheless, these synaptic responses clearly appear as polysynaptic. That polysynaptic responses are suppressed by GABA is not so surprising and does not exclude that axonal branches were first activated to trigger orthodromic responses in the inserted neurons. The lower part of Hirschs Fig. 1 shows suppression of an excivatory postsynaptic current (EPSC) that presents features more typical of a monosynaptic response, which was also suppressed by GABA application. It could be that this EPSC resulted from the direct activation of an initial segment. The small contribution of axon initial segments to the pool of neuronal elements activated to produce orthodromic responses can be understood given the difference in density for axonal branches and initial segments (or cell bodies). The density in cell body of rodent visual cortex is around 100000/mm2 of cortical surface (Peters 1987; Schüz and Palm 1989; Beaulieu 1993). Data for axonal density are not available (to our knowledge) for rat visual cortex, but those obtained for mouse area 17 should give a correct approximation: according to data presented by Braitenberg and Schüz (1991), the density of axonal branches should be 4100000/mm2. In other words, there are 41 times more axonal branches than initial segments in a given volume of cortical gray matter. If initial segments and axonal branches display the same threshold for activation, then the initial segments activation would account for only 2.4% of the postsynaptic response obtained. This is even lower than the 15Ð20% determined from our experiments. This discrepancy can be related to different parameters: first, it could reflect a higher excitability for the initial segments compared to axonal branches. This could be related to the larger diameter of the axon initial segment (Farinas and DeFilipe 1991) and of the main axonal trunk compared with the diameter of axon collaterals, especially the unmyelinated axon collaterals (Peters and Sethares 1996). Second, a more precise evaluation of the contribution of axonal branches compared with that of initial segment might be obtained 499 if the density of the nodes of Ranvier was known, although not all axons are myelinated in the cortical gray matter. Finally, the axon of cortical neurons branches extensively in the vicinity of the cell body as well as in distant patches, such that the axon from a single cell might contribute several times to the number of axons encountered in one sections: this might lead to an overestimation of the true proportion of axons compared with that of cell bodies. Conclusion Whenever a potsynaptic effect is elicited after an electrical stimulation in the cortical gray matter, this effect results mainly from the activation of axonal branches. The biophysical properties of the axon favor their activation by extracellular electrical stimulation (Nowak and Bullier 1997), while the number of axon initial segments is too small with respect to the number of axonal branches to represent a significant component in the responses elicited by electrical stimulation. That electrical stimulation mainly activates axonal branches is not so much a problem if the aim is just to elicit a postsynaptic potential. However, when electrical stimulation is used to activate a given set of connections in order to elicit a response in a particular target, there is a problem that should not be neglected. Examination of the literature shows that, in general, this has not been the case. Electrical stimulation of the cortical gray matter can lead to erroneous interpretation, because, first, the connections linking two points in the cortex are always reciprocal, and second, because all pyramidal cells of the neocortex give rise, in addition to their projection axon, to a number of local axon collaterals arborizing in the vicinity of the cell body. When a stimulation is applied, for example in area 17, it will activate orthodromically efferent axons from this area that project onto area 18. But the stimulation will also induce action potentials in the axons of area 18 neurons that project to or travel near the stimulation site in area 17. The action potentials will propagate antodromically to area 18, where they will invade the intrinsic collaterals. At the end of the collaterals, they will elicit a postsynaptic potential. What happens in that case may sound paradoxical: the resulting postsynaptic response produced in a neuron of area 18 comes, at least partially, from other neurons of area 18, although the stimulation was applied in area 17. Another problem arises when electrical stimulation is used to study the vertical, intracolumnar connections. Responses obtained, for example in layer 3 after electrical stimulation applied in layer 4, result from the activation of axons efferent from layer 4 neurons, but also from the activation of axons that travel through layer 4. These axons have a number of different origins: from other cortical areas, from lower cortical layers, from subcortical structures, and from layer 3 neurons as well. To avoid such problems, the only alternative is to use stimulation methods that activate only the cell body. One is the acti- vation of a single presynaptic neuron with intracellular stimulation. The drawbacks of that technique are that only one neuron is activated and that it is difficult to use in vivo due to its requirement for a good stability. Chemical activation, with glutamate or other excitatory neurotransmitters, appears to be a mode of stimulation that specifically activates cell bodies in large numbers; but the very useful timing of electrical stimulation is lost. In addition, results obtained with chemical stimualation may suffer interpretation problems linked to the phenomenon of depolarization block. Acknowledgements We thank Pascale Giroud and Naura Choulamountri for help during the experiments, also Pierre-Marie Chorrier for help with the mechanics, Christian Urquizar for help with the electronics. Thanks to Guy Chouvet for advice and help with iontophoresis. Thanks to G.E. Loeb for bibliographical references. Thanks to visitors to the Neuroscience meeting (Washinton 1993), to M. Abeles, H.A. Swadlow, W.T. Newsome, and E.J. Tehovnik for their comments and criticisms on this work while it was on progress. This work was supported by H.F.S.P. RG 55/94. L.G.N. was supported by a fellowship from the Ministre de la Recherche et de la Technologie and by a fellowship from the Fyssen foundation. References Aoki C, Venkatesan C, Go C-G, Mong JA, Dawson TM (1994) Cellular and subcellular localization of NMDA-R1 subunit immunoreactivity in the visual cortex of adult and neonaral rats. J Neurosci 14:5202±5222 Armstrong-James M, Welker E, Callahan CA (1993) The contribution of NMDA and non-NMDA receptors to fast and slow transmission of sensory information in the rat SI barrel cortex. J Neurosci 13:2149±2160 Artola A, Singer W (1990) The involvement of N-methyl-d-aspartate receptors in the induction and maintenance of long-term potentiation in rat visual cortex. Eur J Neurosci 2:254±269 Balestrino M, Aitken PG, Somjen GG (1986) The effects of moderate changes of extracellular K+ and Ca2+ on synaptic and neural function in the CA1 region of the hippocampal slice. Brain Res 377:229±239 Beaulieu C (1993) Numerical data on neocortical neurons in adult rat, with specific reference to the GABA population. Brain Res 609:284±292 Benke TA, Jones OT, Collingride GL, Angelides KJ (1993) N-methyl-d-aspartate receptors are clustered and immobilized on dendrites of living cortical neurons. Proc Natl Acad Sci USA 90:7819±7823 Braitenberg V, Schüz A (1991) Anatomy of the cortex. Statistics and geometry. Springer, Berlin Heidelberg New York Choi DW (1987) Ionic dependence of glutamate neurotoxicity. J Neurosci 7:369±379 Christian EP, Dubek FE (1988) Characteristics of local excitatory circuits studied with glutamate microapplications in the CA3 region of rat hippocampal slices. J Neurophysiol 59:90±109 Colbert CM, Johnston D (1996) Axonal action-potential initiation and Na+ channel densities in the soma and axon initial segment of subicular pyramidal neurons. J Neurosci 16:6676±6686 Crunelli V, Forda S, Kelly JS (1984) The reversal potential of excitatory amino acid action on granule cells in the rat dentate gyrus. J Physiol (Lond) 351:327±342 Deisz RA, Fortin G, Zieglgänsberger W (1991) Voltage dependence of excitatory postsynaptic potentials of rat neocortical neurons. J Neurophysiol 65:371±382 Farinas I, DeFilipe J (1991) Pattern of synaptic input on corticocortical and corticothalamic cells in the visual cortex of the cat: the axon initial segment. J Comp Neurol 304:70±77 500 Flatman JA, Schwindt PC, Crill WE, Stafstrom CE (1983) Multiple actions of N-methyl-d-aspartate on cat neocortical neurons in vitro. Brain Res 266:169±173 Garthwaite G, Hajos F, Garthwaite J (1986) Ionic requirements for neurotoxic effects of excitatory amino acid analogues in rat cerebellar cortex. Neuroscience 18:437±447 Gogan P, Gueritaud JP, Tyc-Dumont S (1983) Comparison of antidromic and orthodromic action potentials of identified motor axons in the cats brain stem. J Physiol (Lond) 335:205±220 Hablitz JJ (1982) Conductance changes induced by dl-homocysteic acid and N-methyl-dl-aspartic acid in hippocampal neurons. Brain Res 274:149±153 Hablitz JJ, Lundervold A (1981) Hippocampal excitability and changes in extracellular potassium. Exp Neurol 71:410±420 Hirsch JA (1995) Synaptic integration in layer IV of the ferret striate cortex. J Physiol (Lond) 483:183±199 Huguenard JR, Hamill OP, Prince DA (1988) Developmental changes in Na+ conductances in rat neocortical neurons: appearance of a slowly inactivating component. J Neurophysiol 59:778±795 Huntley GW, Vickers JC, Janssen W, Brose N, Heinemann SF, Morrison JH (1994a) Distribution and synaptic localization of immunocytochemically identified NDMA receptor subunit proteins in sensory-motor and visual cortices of monkey and human. J Neurosci 14:3603±3619 Huntley GW, Vickers JC, Morrison JH (1994b) Cellular and synaptic localization of NMDA and non-NMDA receptor subunits in neocortex: organization features related to cortical circuitry, function and disease. Trends Neurosci 17:536±543 Lehmann A, Jakobson I (1990) Ion dependence and receptor mediation of glutamate toxicity in the immature rat hippocampal slice. Eur J Neurosci 2:620±628 Lipski J (1981) Antidromic activation of neurones as an analytic tool in the study of the central nervous system. J Neurosci Methods 4:1±32 MacDonald JF, Porietis AV, Wojtowicz JM (1982) l-Aspartic acid induces a region of negative slope conductance in the currentvoltage relationship of cultured spinal cord neurons. Brain Res 237:248±253 McNamara D, Dingledine R (1990) Dual effects of glycine on NMDA-induced neurotoxicity in rat cortical culture. J Neurosci 10:3970±3976 Merrill EG, Ainsworth A (1972) Glass-coated platinum-plated tungsten microelectrodes. Med Biol Eng 10:662±672 Nowak L, Bregestovski P, Ascher P, Herbert A, Prochiantz A (1984) Magnesium gates glutamate-activated channels in mouse central neurones. Nature 307:462±465 Nowak LG, Bullier J (1996) Spread of stimulating current in the cortical grey matter of rat visual cortex studied on a new in vitro slice preparation. J Neurosci Methods 67:237Ð248 Nowak LG, Bullier J (1997) Axons, but not cell bodies, are activated by electrical stimulation in cortical grey matter. I. Evidence from chronaxie measurements. Exp Brain Res 118:477±488 Nowak LG, James AC, Bullier J (1997) Corticocortical connections between visual areas 17 and 18a of the rat studied in vitro. Spatial and temporal organisation of functional synaptic responses. Exp Brain Res 117:219±241 Peters A (1987) Number of neurons and synapses in primary visual cortex. In: Jouensand EG, Peters A (eds) Further aspects of cortical function, including hippocampus. (Cerebral cortex, vol 6). Plenum Press, New York, pp 267±294 Peters A, Sethares C (1996) Myelinated axons and the pyramidal cell modules in monkey primary visual cortex. J Comp Neurol 365:232Ð255 Petralia RS, Yokotani N, Wenthold RJ (1994) Light and electron microscope distribution of the NMDA receptor subunit NMDAR1 in the rat nervous system using a selective anti-peptide antibody. J Neurosci 14:667±696 Rothman SM (1985) The neurotoxicity of excitatory amino acids is produced by passive chloride influx. J Neurosci 5:1483±1489 Schüz A, Palm G (1989) Density of neurons and synapses in the cerebral cortex of the mouse. J Comp Neurol 286:442±455 Schwartzkroin PA (1975) Characteristics of CA1 neurons recorded intracellularly in the hippocampal in vitro slice preparation. Brain Res 85:423±436 Stafstrom CE, Schwindt PC, Chubb MC, Crill WE (1985) Properties of persistent sodium conductance and calcium conductance of layer V neurons from cat sensorimotor cortex in vitro. J Neurophysiol 53:153±170 Sutor B, Hablitz JJ (1989a) EPSPs in rat neocortical neurons in vitro. I. Electrophysiological evidence for two distinct EPSPs. J Neurophysiol 61:607±620 Sutor B, Hablitz JJ (1989b) EPSPs in rat neocortical neurons in vitro. II. Involvement of N-methyl-d-aspartate receptors in the generation of EPSPs. J Neurophysiol 61:621±634 Thomson AM (1986a) A magnesium-sensitive post-synaptic potential in rat cerebral cortex resembles neuronal responses to Nmethylaspartate. J Physiol (Lond) 370:531±549 Thomson AM (1986b) Comparison of responses to transmitter candidates at an N-methylasparatate receptor mediated synapse, in slices of rat cerebral cortex. Neuroscience 17:37±47 Thomson AM, Deuchars J (1994) Temporal and spatial properties of local circuits in neocortex. Trends Neurosci 17:119±126 Thomson AM, Gilderstone D, West DC (1989) A local circuit neocortical synapse that operates via both NMDA and non-NMDA receptors. Br J Pharmacol 96:406±408 Thomson AM, Deuchars J, West DC (1993) Single axon excitatory postsynaptic potentials in neocortical interneurones exhibit pronounced paired pulse facilitation. Neuroscience 54:347±360