INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The Hebbian rule for synaptic plasticity emerged from a neuro-computational urge to implement learning and memory function with it (Hebb 1949) and became one of the most influential postulates (Sejnowski 1999) since the discovery of long-term potentiation (LTP) of hippocampal synapses in the early 1970s (Bliss and Collingridge 1993; Bliss and Lømo 1973). Some of the earliest studies of LTP in the hippocampus demonstrated that tetanic activation of presynaptic fibers coincident with accentuated postsynaptic depolarization lead to LTP (Gustafsson et al. 1987; Kelso et al. 1986; Levy and Steward 1979). These and other studies lead to the adoption of LTP as the primary cellular model of Hebbian plasticity (Wigstrom et al. 1988) even though the proposed induction mechanism of LTP was not exactly the same as what Hebb postulated.

Recent findings of spike back-propagation into the dendrite (Spruston et al. 1995) and its involvement in LTP induction (Magee and Johnston 1997; Markram et al. 1997) potentially elevate the phenomenon of LTP to a closer cellular resemblance of Hebb's postulate (Linden 1999). In other words, if dendritic depolarization coincides with tetanus-induced spikes at the soma, LTP appears to be a valid physiological instantiation of the memory mechanism postulated by Hebb. Despite these new discoveries and the realization about how they might relate to Hebb's original theory, depolarization at the dendrite and its temporal pairing with somatic spikes produced during tetanic stimulation has never been directly observed.

To examine this pairing, it is important to simultaneously measure the spatial and temporal coincidence of the membrane potential response to an afferent tetanic stimulus at multiple sites along dendritic processes. Optical imaging of neuronal activities with a voltage-sensitive dye (VSD) (see reviews by Cohen et al. 1978; Grinvald et al. 1988; Ichikawa et al. 1993) is one potentially suitable method to examine simultaneously the excitation of postsynaptic neurons at multiple sites. We recently established a stable and reliable measurement system and quantified the optical signals of a fluorescent type of VSD aminonaphthylethenylpyridinium (Di-4-ANEPPS) (Loew et al. 1992; Yuste et al. 1997) in the rat hippocampal slice preparation with corresponding electrophysiological parameters. A major conclusion was that the optical signals largely reflect the population activity of postsynaptic principal neurons (Tominaga et al. 2000b). In the present study, we took advantage of the simultaneous multi-site recording afforded by the optical recording method with Di-4-ANEPPS to examine the neuronal response to tetanic stimulation and to evaluate whether and to what extent Hebb's postulates are precisely applicable to the induction process of tetanus-induced LTP.

It should be noted that the optical imaging of VSD at the tissue level only allowed us to observe the signal produced by a population of neurons. That is, 1 pixel of the imaging device collected the optical signal in the light path of a tissue segment of about 22 × 22-µm square that contains processes of many neurons. However, because the principal neurons of the hippocampal slice are arranged in highly discreet lamina, this enabled us to assign the population signals to particular regions of pyramidal neurons. Hence we can extrapolate this population signal to a summed single-cell level and can examine the coincidence of the membrane potential response at the dendrites and soma. However, the absolute calibration of optical signals in terms of membrane potential change is not possible as described earlier. Thus the conclusions derived from signal amplitudes are based on comparing signals from the same region of a slice under different experimental conditions or from different locations that belong to the same anatomical layer (with uniform anatomy).

This experimental approach also enabled us to examine the circuit behavior in area CA1 after afferent burst stimulation. Because we can assume a similar sensitivity of the VSD to the actual membrane potential response in the same layers, we can examine the degree of activity spread within different locations in a given layer. For example, we could evaluate how widely the optically detected depolarization would spread along the transverse axis (e.g., proximal CA1 to distal CA1) in response to a standard tetanic stimulation compared with that after a single stimulation or other types of burst stimulation. These kinds of comparisons of the spatial and temporal structure of the optically detected long-lasting depolarization revealed novel aspects of temporal and spatial activity of CA1 intrinsic circuitry in response to burst stimulation.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Slice preparation and staining with VSD

Hippocampal slices were prepared from 4- to 5-wk-old male rats decapitated under deep ether anesthesia. The brains were quickly cooled in iced artificial cerebrospinal fluid [ACSF, which contained in (mM) 124 NaCl, 2.5 KCl, 2 CaCl₂, 2 MgSO₄, 1.25 NaH₂PO₄, 26 NaHCO₃, and 10 glucose, pH 7.4 after bubbling with mixed 95% O₂-5% CO₂ gas]. After cooling for 5 min, the hippocampus was dissected out along with the surrounding cortex and sliced into 400-µm-thick sections with a vibratome (752 M Vibroslice, Campden Instruments, Loughborough, UK). Each slice was transferred onto a fine-mesh membrane filter (Omni Pore membrane filter, JHWP01300, Millipore) held in place by a thin Plexiglas ring (11 mm ID; 15 mm OD; 1- to 2-mm thickness) after a short incubation in 95% O_2-5% CO₂-saturated ACSF for 3-5 min. Slices placed in the Plexiglas ring were transferred to a moist holding chamber continuously supplied with a moistened mixture of the O₂ and CO₂ gas mixture. The temperature of this chamber was held at 32°C for 1 h and then at room temperature thereafter. After 1 h of incubation in this chamber, slices were stained for 25 min with an aliquot of the VSD staining solution (100 µl for each slice). This solution consisted of 0.2 mM Di-4-ANEPPS (D-1199, Molecular Probes) in 2.7% ethanol, 0.13% Cremophor EL (Sigma), 50% fetal bovine serum (Sigma), and 50% ACSF. The slices were subjected to experiments after at least 1 h incubation in normal ACSF at room temperature.

Recording

For recording, a slice supported by the Plexiglas ring was transferred to an immersion-type recording chamber. Slices were continuously perfused with ACSF at a rate of 1 ml/min. The ACSF was continuously bubbled with a 95% O₂-5% CO₂ gas mixture and warmed to 31°C with an electronic temperature controlling device (PSMI module, Medical Systems) before being channeled to the recording chamber. Laboratory-designed epifluorescence optics, consisting of two principal lenses (f = 50 mm F/1.4 Nikon objective lens and ×1.0 Leica Microsystems projection lens), a dichroic mirror (575 nm) and absorption (530 nm) and excitation (590 nm) filters, were mounted above the slice. Emitted fluorescence was collected and projected onto a laboratory-designed CCD-camera (commercially available through BrainVision as MiCAM01). The ratio of the fractional change in fluorescence of VSD to the initial, prestimulation amount of fluorescence (F/F) was calculated and used as the optical signal. The optical signals presented in the following sections are filtered in spatial and temporal domains using a Gaussian kernel of 5 × 5 × 3 (horizontal × vertical × temporal directions). In some experiments, when a drift of the baseline was observed, it was compensated for by subtracting a normalized smooth spline curve obtained from optical signals recorded at pixels where no response was observed (e.g., optical signals in the hilus). We confirmed that this procedure gave steady and flat baselines and did not cause any artificial drift in signals when no stimulus was applied. The analyses of the optical signals were done with a procedure developed for Igor Pro (WaveMetrics).

A glass microcapillary tube (5 µm OD, filled with ACSF) was used as a monopolar stimulating electrode and a recording electrode for field potential recordings. For intracellular recording, we used a fine-tipped glass microelectrode filled with 4 M potassium acetate (approximately 100 M). An Axoclamp-2B amplifier (Axon Instruments) was used in continuous bridge mode. Cells with resting potentials of 65 to 80 mV were accepted for study. The electrophysiological recording system was controlled with a procedure developed for Igor Pro (WaveMetrics). Part of the program used PULSE XOPS (Herrington et al. 1995). Typically, a 0.05- to 0.1-Hz stimulus, with an intensity that produced an approximately 30% maximal field excitatory postsynaptic potential (fEPSP), was applied to the Schaffer collateral pathway in most of the experimental procedures to monitor synaptic transmission. Neither of these experimental processes interfered with the other. For details regarding the optical recording technique, see our previous paper (Tominaga et al. 2000b).

D,L-threo--hydroxyaspartate (THA, Sigma) and D,L-2-amino-5-phosphonovaleric acid (D,L-APV, Tocris) and other common reagents were obtained through local resellers.

Statistical significance was determined by one-way ANOVA, paired Dunnett's tests using the Igor Pro software (WaveMetrics).

Some of these results were reported in preliminary form elsewhere (Tominaga et al. 1999, 2000a).

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

LTP-inducing tetanic stimulation elicits sustained depolarization in each lamina of area CA1

We imaged evoked neural activity in area CA1 of rat hippocampal slices using the VSD Di-4-ANEPPS and optically recorded its fluorescence with a high-speed camera (0.7 ms/frame; 90 × 60 pixel resolution). Figure 1 shows representative tetanus-induced changes in the electrophysiological and optical signals accompanying the induction of synaptic LTP in area CA1 of rat hippocampal slices. The rectangle in the inset of Fig. 1A showing the placement of electrodes represents the camera's field of view over the hippocampal slice. Evoked changes in membrane potential after tetanic stimulation of the Schaffer collateral/commissural pathway were represented as the fractional change of fluorescence in the optical signal. Standard intracellular recordings of putative pyramidal cells (Fig. 1A, inset; intra.) and standard field potential recordings in stratum radiatum were made simultaneously with the optical recordings (Fig. 1A, inset; f.p.). The change in amplitude of the fEPSP slope over time is shown in Fig. 1A. Traces of membrane potential change recorded by an intracellular electrode and field potential change on a tetanic stimulation are shown in Fig. 1, B and C, respectively. Corresponding, simultaneously recorded optical signals recorded at a pixel close to the electrode tips of each electrode are shown in Fig. 1D (pyr. and rad.). The optical signal for each time sample is shown in Fig. 1E.

View larger version (48K): [in this window] [in a new window]   Fig. 1. Tetanus-induced changes in the optical and electrophysiological signals accompanying the induction of synaptic long-term potentiation (LTP) in area CA1 of rat hippocampal slices. A: representative changes in the amplitude of the field excitatory postsynaptic potential (fEPSP) slope after tetanic stimulation (100-Hz, 1-s duration, 40 µA) of the Schaffer collateral/commissural pathway. Inset: the placement of the glass stimulating electrode (Stim) in s. radiatum at the CA1/CA2 border, the fine-tipped intracellular recording electrode (Intra) in s. pyramidale of CA1, and the glass extracellular electrode for monitoring field potentials (f.p.) in s. radiatum of CA1 near the intracellular electrode. The rectangle superimposed over the slice represents the imaging device's field of view. B: intracellular recording of the tetanus-induced membrane potential change. C: the field potential change induced by the tetanus. D: simultaneously recorded optical signals (1.5 ms/frame, 2,730 frames, 1 scan). Top (pyr): the response obtained from a pixel (22.2 × 21.7 µm in the objective plane) close to the cell in which the intracellular electrode was blindly inserted. Bottom (rad): the response obtained from a pixel close to the tip of the extracellular electrode. E: consecutive pseudocolored images (each image is 2 × 1.3 mm in the objective plane and consists of 90 × 60 pixels) showing the 1st 1.8 s of stimulation-evoked changes in the optical signal spreading over area CA1. These optical voltage maps were superimposed on the initial, prestimulation gray-scale fluorescent image of area CA1. The numbers in each image show the time(s) after the start of tetanic stimulation. Stimulation artifacts in the electrophysiological traces of this figure (B and C) and subsequent ones were removed digitally.

Tetanic stimulation of the afferent fibers elicited LTP in the fEPSP (Fig. 1A), sustained depolarization of the pyramidal cell membrane (Fig. 1B), and a sustained negative shift in the field potential in s. radiatum (Fig. 1C). The time course of the optical signals obtained from an individual pixel close to the recorded pyramidal cell (Fig. 1D, top) and from a pixel in s. radiatum (Fig. 1D, bottom) was essentially identical to that of the electrical signal. The depolarizing signal spread throughout area CA1 along the Schaffer/commissural pathway (Fig. 1E).

Several tissue elements may have contributed to the optically detected long-lasting depolarization besides CA1 pyramidal cells. To examine the contribution of the glial membrane potential change, we applied D,L-threo--hydroxyaspartate (THA, 1 mM), which is known to reduce the glial components in the optical signal when an absorption type of VSD (RH-155) is used (Kojima et al. 1999). Figure 2A shows the effect of THA bath application on the time course of optical signals obtained from a representative pixel (middle of s. radiatum) in the control condition (black trace) and after application of THA (blue trace). As shown in the figure, bath application of THA failed to reduce the response much. The small, initial decrease in the response might be caused by reduced synaptic transmission observed in the fEPSP after single stimulations. It is interesting to note that THA increased the decay time of the response.

View larger version (20K): [in this window] [in a new window]   Fig. 2. Contribution of glial depolarization and presynaptic axonal depolarizations to the optically recorded LTP. A: representative traces of the optical signals obtained from a representative pixel (middle of s. radiatum) in the control condition (black trace) and after application of D,L-threo--hydroxyaspartate (THA, blue trace), a glial glutamate transporter blocker. B: representative traces of the optical signals obtained from a representative pixel (middle of s. radiatum) in the control condition (black trace) and after application of low-Ca2+ solution (blue trace). B is an averaged trace of 4 trials.

We next examined the contribution of the presynaptic component to the optically recorded depolarization by perfusing low-Ca²⁺ medium (0.25 mM Ca²⁺, 6 mM Mg²⁺) into the recording chamber; the other components of the ACSF remained the same. Figure 2B shows the effect of low Ca²⁺ in the medium on the time course of the optical signal obtained from a representative pixel in the middle of s. radiatum (control condition, black trace; low-Ca²⁺ medium, blue). As is clear in the figure, the long-lasting depolarization was almost completely suppressed except for focal small signals in the vicinity of stimulating electrode, which correspond to direct depolarizations of membranes by individual stimulations. Taken together, these observations suggest that the long-lasting depolarizing signal is primarily caused by the postsynaptic pyramidal neurons. Also supporting this conclusion are the observations that the time course of the optical signal and intracellular recording from the pyramidal neuron were very similar (Fig. 1B), and the distribution of interneurons is relatively sparse (only ca. 11% of the neurons are GABAergic) (Woodson et al. 1989).

N-methyl-D-aspartate receptors do not contribute to the sustained depolarization

Tetanus is believed to induce depolarization at the dendritic membrane, which if strong enough, removes Mg²⁺ions from the N-methyl-D-aspartate (NMDA)-receptor channel in a voltage-dependent manner (Nowak et al. 1984). This channel unblocking, in turn, allows Ca²⁺influx that activates a signal cascade of events that ultimately leads to the induction of LTP (Bliss and Collingridge 1993). To estimate the contribution of the NMDA receptor current and LTP-induced postsynaptic modifications on sustained depolarization (produced during afferent tetanic stimulation), we examined the effects of both 50 µM D,L-APV, an NMDA-receptor antagonist, and LTP. The effects of LTP on sustained depolarization was examined both during induction and after LTP was established.

Figure 3A shows representative changes in fEPSP slope amplitudes on four successive tetanic stimulations (100 Hz over 1 s each with interval more than 3 min) in the presence of APV (i-iv APV), after washout of APV (i-iv after wash) and after establishment of LTP (i-iv LTP). In Fig. 3B, mean amplitudes of the optically detected sustained depolarization at the tip of the extracellular electrode for each episode of tetanus in Fig. 3A are shown (mean and SE, n ranges from 3 to 6). In the presence of D,L-APV, none of four successive tetanic stimuli (i-iv in Fig. 3A) caused a persistent change in the fEPSP slope (Fig. 3A, left), while at the same time produced normal sustained depolarization (APV in Fig. 3B). After removing D,L-APV, the first tetanus (i, Fig. 3A, middle) of the next set of four stimuli enhanced the fEPSP, but the amplitude of the associated sustained depolarization was not significantly different from that depolarization associated with the first tetanus when D,L-APV was present (cf. Fig. 3A, i in APV vs. i in after wash and B, i in APV vs. i in after wash). This suggests that the NMDA-receptor current contributes less to sustained depolarization. This might be due to a less obvious reversal potential or a small conductance. However, we noticed a slight effect of APV near the stimulating electrode, which corresponds to the finding of Herron et al. (1986).

View larger version (39K): [in this window] [in a new window]   Fig. 3. Effect of 2-amino-5-phosphonovaleric acid (APV) and the induction and establishment of LTP on sustained tetani-produced depolarization. A: changes in amplitude of fEPSP slope before, during, and after tetanic stimulation. Four bouts of tetani (100 Hz over 1 s each) were applied in the presence of 50 µM APV (i-iv in APV) after the establishment of a stable baseline response. Following this set of stimuli and APV washout, a 2nd set of 4 tetani (i-iv in after wash) was applied and a 20-min delay was imposed, during which LTP became established. Finally a 3rd set of 4 tetani (i-iv in LTP) was applied 30 min after the end of the 2nd set when LTP was clearly present. B: mean amplitude of the optically detected sustained depolarization at the tip of the extracellular electrode for each episode of tetanus shown in A. The error bars show the SE n ranges from 3 to 6 experiments. Statistically significant differences between response to the initial tetanus (i in APV) and subsequent responses are indicated with asterisks (***P < 0.01, **P < 0.05).

Subsequent tetani in this set (ii-iv) produced a larger response (Fig. 3B, *). After stable LTP was induced, a third set of tetani tended to produce a larger response (Fig. 3B, LTP), but these sustained depolarizations were unreliable and not significantly different from the first one of the set. This latter result suggests some involvement of LTP-related postsynaptic activity in producing sustained depolarization.

Because the application of APV did not alter the long-lasting depolarization but inhibited the induction of LTP, we always applied APV in the following experiments to examine the long-lasting depolarization in isolation.

Sustained depolarization breaks electrotonic conduction from dendrite to soma and causes spike failure

Except for a few initial spikes, the apparent absence of an action potential during tetanus in the intracellular recordings is notable (Fig. 1B). To examine this in the optical signals with a better signal-to-noise (S/N) ratio, we averaged optical responses from four successive trials with intervals longer than 3 min in the presence of D,L-APV. In Fig. 4, A and B, traces of the averaged optical signals obtained from a pixel near the stimulating electrode during tetanic stimulation (red) and a single stimulus (black) are shown. The top set of traces is from s. pyramidale, and the bottom traces are from s. radiatum. Figure 4, A and B, shows that there were only a few spike-like rapid transients in the optical signals collected from s. pyramidale (top) at the beginning of the tetanic stimulus, while a repeated response to each stimulus was seen in s. radiatum (bottom). With respect to their time courses (see Fig. 4, B and E), the sharp signals observed at s. pyramidale most likely correspond to the population signal of the action potentials, some of which are seen in the intracelluar recording (Figs. 1B and 4C); the preceding, repeating responses observed at s. radiatum may reflect mostly EPSPs at the dendrites (for a discussion regarding their amplitude, see DISCUSSION). These responses were "riding on top of" a slow sustained depolarizing signal. That is, the pairing of the dendritic responses with spike generation at soma was practically never prominent during a tetanic stimulation. In fact, the pairing was only seen during the first few stimuli in tetanic stimulations. Virtually wherever long-lasting depolarizations could be elicited in CA1, a similar absence of pairing was observed.

View larger version (56K): [in this window] [in a new window]   Fig. 4. A lack of spikes and a decrease in the membrane resistance of a CA1 principal neuron during tetanic stimulation. A: the time course of optical signals obtained from a pixel near the stimulating electrode during tetanic stimulation and a single stimulus. The red traces show the optical response (0.7 ms/frame, 2,730 frames, an average of 4 with an interval of at least 3 min) to tetanic stimulation (100-Hz, 400-ms duration, 200 µA stimulation), while the black traces show the response (0.7 ms/frame, 341 frames, 16 average with interval of 20 s) to a single stimulus with the same stimulus intensity. The top set of traces are signals obtained from s. pyramidal and the bottom traces are those obtained from stratum radiatum (approximately 200 µm distant from s. pyramidale), where the signal strength, and thus depolarization, was maximal. B: portions of the same traces as in A but expanded in the time (10×) and fluorescence intensity (5×) scales. Tick marks below the lower traces represent the tetani. C: time course for the recovery of input resistance of the pyramidal cell after cessation of tetanic stimulation. The resistance was determined from the membrane potential shift caused by the injection of a 4-nA square current pulse lasting 40 ms that was applied to the cell every 80 ms (trace). D: time course for the optical signal associated with the membrane potential change in distal CA1 along the axis orthogonal to the pyramidal cell layer (shown schematically at the left of the figure with the corresponding gray-scaled fluorescence image of a strip; the strip corresponds to an area surrounded by blue-line in the right-hand fluorescent image superimposed on a illustration of a hippocampal slice). D and E, inset: a fluorescent image superimposed on an illustration of a hippocampal slice. Red dots in the rectangle show position of pixels corresponding to the traces shown in A and B. Red dot near the CA3/CA1 border shows the position of stimulating electrode tip. SO-A, s. oriens-alveus; SP, s. pyramidale; SR, s. radiatum; SL-M, s. lacunosum-moleculare. F: profiles of optical signals shown in D and E are plotted for the maximum response during the time periods 0-10 ms (i), 20-30 ms (thick red line) (ii), 100-110 ms (iii), 200-210 ms (iv), and 400-410 ms (v) from the start of tetanic stimulation (arrowheads in the D and E). The profiles of i-v were normalized by its maximum amplitude and plotted in the right-hand graph. A strip of fluorescent image and illustration of pyramidal cells are placed in the most right-hand column. Zero on the ordinate corresponds to the lower edge of the s. pyramidale identified in a prestimulation fluorescent image of the slice.

The fact that most repeated responses occurring in the dendrites failed to elicit somatic potential changes, suggests that some change in the passive membrane electrical properties of the pyramidal cell occurs during tetanic stimulation. To examine this directly, we measured the membrane conductance change occurring during tetanic stimulation by applying 4-nA square current pulses (40-ms duration) through the intracellular electrode. A representative intracellular recording is shown in Fig. 4C. Calculated changes in membrane resistance were fitted to a single exponential function and plotted (Fig. 4C, bottom) with the same time scale as that in the top trace (dotted line indicates the resting membrane resistance calculated from a potential shift just before the application of a tetanic stimulation). Figure 4C clearly demonstrates a change in the passive membrane electrical properties, showing a 23.3 ± 8.9% (resting input impedance, 45.9 ± 18.1 M; minimum input impedance, 10.3 ± 4.5 M; n = 8) reduction in the resting membrane resistance of the pyramidal cell, which recovers exponentially with time (time constant, 1.6 ± 0.36 s, n = 8).

Figure 4, D and E, shows the time course of the optical signal in a strip along the long axis of the pyramidal cells in pseudocolor. This is shown schematically at the left of the figure with the corresponding gray-scaled fluorescence image of a strip, which corresponds to an area surrounded by the blue-line in the right-hand fluorescent image superimposed on a illustration of a hippocampal slice. The time courses correspond to the traces in Fig. 4, A and B. Depolarizing optical signals were seen along almost the whole radial extent of CA1 (i.e., through s. oriens-alveus to s. lacnosum-moleculare) on the first two stimulations. In contrast, the subsequent stimulus elicited little response at s. pyramidale and s. oriens-alveus, whereas in s. radiatum each stimulation of the tetanic burst elicited responses. In Fig. 4F (left), the profiles of the optical signals associated with neural activity at each time period, i-v, were plotted against the distance from the lower edge of the s. pyramidale toward the hippocampal fissure. It should be noted that the profile does not represent the true membrane potential profile along the pyramidal cells. However, it is worth comparing amplitude profiles of optical signals, especially those elicited by single stimulation, because of much accumulated electrophysiological evidence about the membrane potential response to a single stimulation recorded with traditional methods. The time period i (i.e., red thin profile on the plot) represents the response to the first stimulus of the tetanic stimulation, so it should be identical to the response to a single stimulation applied in isolation. The profile of the optical signal to the third stimulus of tetanic stimulation (time period ii) showed a similar amount of activation at s. radiatum compared with period i, but the response beyond s. pyramidale was much smaller than that of the time period i (Fig. 4F and also see E). The 10th stimulus of the tetanic stimulation (100 ms from the beginning; time period iii) elicited a greater optical response at s. radiatum, but a smaller optical response at s. pyramidale. As the number of stimulations increased, the amplitude of the optical signal at s. radiatum increased, and that beyond s. pyramidale also increased. The latter increase in optical signal beyond s. pyramidale might be due simply to electrotonic spread. To show more clearly this possibility, we normalized each profile by its maximum value to compare the shapes of profiles, as shown in Fig. 4F, right. Comparing the response to a single stimulation (i) to the others in the series, normalized profiles after 20 ms reflect deceased relative activation at s. pyramidale and similar shapes. This suggests that the optical signal at s. pyramidale after 20 ms increased almost proportionally to the increase in the optical signal at the s. radiatum. It implies that the activation of the pyramidale cell layer is decreased after 20 ms of stimulation and most possibly that the optical signal seen there is a passive electrotonic component.

Response profile of the optical signals along the transverse axis of area CA1 associated with long-lasting depolarization from tetanic stimulation is steeper than the response to single stimulation

Figure 5A shows maximum amplitude of the response in three-dimensional (3-D) format after a single stimulus (top) and after a standard tetanic stimulation (bottom). The figure represents an overall profile of the optically detected response and does not exactly match the actual potential profile of each neuron. The portion of the pyramidal cell dendrites where the depolarizing optical signal was maximal in response to a single stimulus seems to be the same, irrespective of the transverse position in CA1, because the "ridgeline" (i.e., maximum elevation) of the response in the 3-D plot parallels s. pyramidale (Fig. 5A, top). This was also the case for the sustained depolarizing optical response to tetanic stimulation. After tetanic stimulation, the distance from this ridgeline to s. pyramidale in the plot was 233.4 ± 33.6 µm (n = 6), which was not significantly different (P > 0.05) from the corresponding distance (223.4 ± 47.5 µm; n = 6) for a single stimulus-elicited response (Fig. 5A, top).

View larger version (35K): [in this window] [in a new window]   Fig. 5. Progressive convergence of the response along the Schaffer collaterals in CA1 on standard tetanic stimulation (100-Hz, 400-ms duration, 200 µA) and a broader response on -burst stimulation (15 sets of 4 100-Hz stimulations with an interval of 200 ms, 200 µA). A: 3-dimensional presentation of the maximum amplitude of the response after a single stimulus (top) and after a standard tetanic stimulus (bottom). B: normalized spatial profiles for the maximum responses (along the "ridge line" of the response parallel to the s. pyramidal) during the time periods 0-10 ms (i), 100-110 ms (ii), 200-210 ms (iii), 300-310 ms (iv), and 400-410 ms (v). C: individual responses to each tetanic stimulus (bottom) were isolated by first fitting the raw response to a single time exponential (time constant = 200 ms, red line in top trace) and then subtracting this fit from the raw response. D: normalized spatial profile of the -burst-induced response. E: a similar procedure to that in C was used to isolate each response for each -burst stimulus.

It was notable that the long-lasting depolarization after tetanic stimulation was more prominent around the stimulating electrode compared with the response after a single stimulation. To examine this more closely, we plotted the profile of the amplitude of the long-lasting depolarization along the ridgeline of the response as described in the preceding text. The normalized spatial profile for the maximum response (along the transverse axis of hippocampal slice at the level of "ridgeline" of the response) is shown in Fig. 5B for each time period (0-10 ms, i; 100-110 ms, ii; 200-210 ms, iii; 300-310 ms, iv; 400-410 ms, v). As shown in Fig. 5B, the profile of the response after the first stimulus of a tetani (i) was relatively constant along the transverse axis of CA1. This response profile to the first stimulus of a tetanus was identical to the profile obtained from a single-stimulation protocol. In contrast, the profile became steeper after 100 ms of tetanic stimulation (ii), and the steepness progressively increased with time (iii-v). This nonlinear decrement in the steepness of the response profiles seems to suggest a decrease in excitation conduction in presynaptic fibers. Figure 5C shows a representative trace obtained at a representative pixel in s. radiatum (top). By subtracting a baseline (i.e., red line in the top trace in Fig. 5C; a single exponential fit of long-lasting depolarization), the response to each stimulus in a tetanic stimulation was isolated (Fig. 5C, bottom). In fact, the amplitude of the individual responses superimposed on the tetanus-induced slow depolarization decreased with time (Fig. 5C).

-burst stimulation caused broad sustained depolarization and accentuated pairing

To test if other types of tetanic stimulation cause the same type of long-lasting depolarization, we imaged the optical response to -burst stimulation (Larson and Lynch 1986) applied to the Schaffer collaterals. This stimulation protocol mimics spontaneous neural activity observed in vivo (Buzsaki et al. 1983). The response induced by -burst stimulation caused sustained depolarization with a broader spatial profile along the pyramidal cell dendrites (Fig. 5D). The reliability of this difference in response profiles resulting from a standard tetanus protocol compared with the -burst stimulation protocol was examined statistically by measuring the rate of the response reduction along the ridgeline in the two kinds of profiles. The relative amplitude 440 µm from the stimulated portion was 63.3 ± 0.3% for the standard tetanus-induced response and 73.6 ± 0.4% for -burst-induced response, which was statistically significant (n = 6, P < 0.05). It is also interesting to note that individual responses to -burst stimulation did not decrease with time (Fig. 5E) as was the case with the responses to a standard tetanic stimulation (cf. Fig. 5. C and E).

In addition, with -burst stimulation, exaggerated pairing of the potential responses was observed in s. pyramidale and s. radiatum. Figure 6A shows representative traces of optical signals acquired at representative pixels in s. pyramidal (top) and s. radiatum (bottom). As is clear in the top trace of Fig. 6A, spike firing was not apparent with the first -burst stimulation, but it became clearer in response to the following bursts (compare the response indicated by arrows). Figure 6, B and C, shows representative intracellular recordings and the fEPSP, respectively, during -burst stimulation. As is most clear in the enlarged figure in Fig. 6, D-F, spike generation at the soma increased with time during -burst stimulation.

View larger version (25K): [in this window] [in a new window]   Fig. 6. Accentuated spike generation and pairing with fEPSP in response to a -burst stimulation. A: representative time courses of the -burst-induced sustained depolarization recorded optically (1.5 ms/frame, 2,730 frames, 1 trial) at s. pyramidale (top) and s. radiatum (bottom). B: representative time course of intracellular recordings obtained from a soma of pyramidal cell. C: representative time course of field potential obtained from s. radiatum. D-F: are the same time trace in A-C in enlarged time scale.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

In the present study, we demonstrated with optical recording a long-lasting depolarization in response to LTP-inducing tetanic stimulation. This was expressed as a fractional change in fluorescence emitted from VSD that is incorporated into the membrane of neurons in the hippocampal slice. The imaging method successfully visualized the population neural activity in each membrane fraction (i.e., basilar dendrites, soma, and apical dendrites) throughout area CA1 of the rat hippocampal slice preparation. Even though the distribution of the true membrane potential change cannot be addressed, the optically detected neural activity lead us to the conclusion that a standard tetanus (100 Hz, 1 s) did not cause accentuated spike generation during tetanic stimulation.

Implication of the optical signals

The time course of the optically observed long-lasting depolarization was largely consistent with that of the intracellular recordings of postsynaptic pyramidal cells, although a few spikes recorded at the cessation of tetanic stimulation, which were presumably not synchronized with spikes from cells in the population (Fig. 1B), were absent in the optical signal (Fig. 1D). A few spikes at the start of tetanic stimulation, which presumably were synchronized, can be seen in the optical signal (Fig. 1D). The apparent discrepancy between the intracellular recordings and optical signals can be explained by the fact that the optical signals reflect the population activity of membrane potential changes, so individual responses that are not synchronized with similar responses from cells of the population cannot be recorded with this method. Knowing the average or population activity of the cells can be advantageous but, at the same time, can be disadvantageous if one's goal is to resolve single or statistically rare signals. Note that the time resolution of the system (0.7 ms/frame) is sufficient to follow fast events, such as spikes, if they occur synchronously, as demonstrated in Figs. 1 and 4.

Application of a low-Ca²⁺ solution reduced the optically detected long-lasting depolarization (Fig. 2B). The VSD used in the present study is known to preferentially report the membrane potential response of postsynaptic neurons rather than glial membrane responses (Tominaga et al. 2000b). Moreover, in the present study, potentially contributing glial membrane voltage changes were specifically ruled out by showing that THA, a glial glutamate transporter blocker, did not alter our signals (Fig. 2A). These results imply that the optical signal detected in each lamina of CA1 on tetanic stimulation primarily reflects population changes in intracellular membrane potential of the membranes of CA1 pyramidal neurons, from the basal dendrites to apical dendrites.

It should be noted that the optical signal of the VSD still does not calibrate in terms of actual membrane potential of the neurons because of several factors, including the difference in the ratio of the active dye, the dye bound to membranes that exhibit membrane potential change, to inactive dye, and the dye bound to membranes that does not exhibit change. Based on these factors, the optical signal cannot be used as an indicator for the absolute amount of neural activity, but can be used to compare the relative magnitude of the response at given pixels or areas across different experimental conditions.

The inability to calibrate optical signals in terms of membrane potential is apparent in the amplitude profile of the optical signal (Fig. 4F). We interpreted the sharp deflection appearing in the optical signal from the s. pyramidale as population action potentials and sharp repetitive optical signals from the s. radiatum as population EPSPs. Even with the first response on tetanic stimulation, which corresponds to the response to a single stimulation, spike-like sharp optical signals obtained from the s. pyramidale (Fig. 4B, top) were smaller than that obtained from s. radiatum (Fig. 4B, bottom; see also the profile shown in F, red thin line). Because of the inability of optical signal calibration and the nature of population signals, we cannot determine whether this amplitude profile reflects the actual amplitude profile in the pyramidal cells. This uncertainty should be addressed by future experiments specifically designed to answer this question.

In Fig. 5A we showed the overall profile of the response to a single stimulation and to a tetanic stimulation. For both cases, the "ridge lines" of the response profiles ran parallel to the transverse axis of the slice, and the distances between these lines to s. pyramidale (about 220 µm) were not significantly different for the two kinds of stimulation. It cannot be concluded that the depolarization was most significant at the level, or the region of the dendrite, because of the relative nature of the VSD signal. However, it is probable that the same region of pyramidal cell dendrites preferentially elicits responses to both single and tetanic stimulation because of presumed uniformity of the tissue in s. radiatum.

Tetanic stimulation (100 Hz) induces non-Hebbian plasticity, whereas -burst stimulation induces Hebbian plasticity

In the present report, we demonstrated the absence of CA1 postsynaptic neuron discharge during standard, 100-Hz tetanic afferent stimulation (Fig. 1). The absence was observed in optical signals (Fig. 4, A and B) recorded from most parts of CA1 and was corroborated in the intracellular recordings (Fig. 1B). This absence of firing can be explained by a blockage of electrical conduction in the pyramidal cell dendrites resulting from a large reduction of membrane resistance accompanying the long-lasting depolarization (Fig. 4C). That is, if this reduction of membrane resistance was to occur, the effect would be to temporarily reduce the membrane length constant of pyramidal cells, which in turn would constrain the membrane potential change to a certain position on the dendrite. The optically detected long-lasting membrane depolarization confirmed that the optically detected depolarization was constrained to the certain portion of the dendrite relative to that elicited by a single stimulation (Fig. 4F). Therefore we concluded that sustained depolarization occurs in pyramidal cell dendrites and this depolarization cannot elicit spikes at the soma due to the decreased membrane resistance that would also lead to a decrease in the cell membrane length constant. In other words, standard tetanus induces LTP in the absence of correlated pre- and postsynaptic firing. Note that the observed absence of firing occurred in the part of CA1 where the preference for LTP-induction is usually greater (Tominaga et al. 2000b) than the position near the stimulating site. The precise preference for LTP-induction along the transverse axis of CA1 will appear in a future report.

LTP of hippocampal synapses has long been thought to be the prime physiological instantiation of Hebb's "memory" postulate, especially because the recognition of spike back-propagation (Spruston et al. 1995) and its role in LTP induction (Markram et al. 1997) through the enhancement of an associated Ca²⁺ signal (Magee and Johnston 1997; Nakamura et al. 1999).

According to our observations, the standard tetanus-induced LTP that has been studied for decades in the hippocampus is suggested to be non-Hebbian in the strict interpretation of the original theory. In contrast, we found that with a -burst afferent tetanus, stimulation that more closely resembles the firing patterns observed in behaving rats, spike firing was enhanced (Figs. 5D and 6). Therefore the -burst-induced LTP is suggested to be Hebbian, as the postulate was originally formulated.

Although our observation suggested that the standard tetanus-induced LTP is less likely to elicit a Hebbian type of LTP (as it was originally postulated), it would be still possible that augmented activity of some kind of voltage-sensitive channels in the dendrite could participate in the induction of LTP due to elevated long-lasting membrane depolarization at the dendrite (i.e., a local "Hebbian" rule).

Steeper spatial convergence of the sustained response along the transverse axis of CA1 suggests conduction block in presynaptic fibers

The optical signal profiles of the sustained depolarization response along the ridgeline in 3-D spatial plots of the CA1 potential converged at the stimulation site more steeply and prominently than the response profiles to single stimulation (Fig. 5A, bottom). Because we can presume a uniform synaptic strength irrespective of distance, the amplitude of the response at the ridgeline could be used to estimate the number of presynaptic fibers that successfully transmit a presynaptic action potential to pyramidal cells. Hence, the convergent response can be explained by conduction blockage of presynaptic fibers in the region where sustained depolarization is elicited.

Why is this conduction blockage in the presynaptic fibers accompanied by sustained depolarization? We propose that presynaptic fibers also depolarize when tetanus produces sustained depolarization of pyramidal cells, in the following manner (Fig. 7). 1) Transmitter release from presynaptic fibers causes sustained depolarization in the principal neuron. 2) The sustained depolarization of the principal neuron or intense activation of interneurons on tetanic stimulation elevates extracellular K⁺ (Smirnov et al. 1999; Taira et al. 1997). 3) This build-up causes sustained depolarization in the presynaptic fiber and should result in sustained depolarization of the principal neuron. 4) Consequently, conduction blockage occurs at the depolarized portion of the fiber. It is probable that tetanic stimulation induced long-lasting depolarization is caused partly via intense activation of GABA_A receptors. Several studies indicate that intense activation of GABA_A receptors, sometimes by a brief application of high-frequency stimulation to the presynaptic fibers, causes a depolarizing inhibitory postsynaptic potential (Alger and Nicoll 1979; Grover et al. 1993; Staley and Proctor 1999).

View larger version (40K): [in this window] [in a new window]   Fig. 7. A schematic illustration of the proposed mechanism of the conduction block of presynaptic fibers from accumulation of extracellular K+ (green patch) produced by accentuated activation of principal neurons and interneurons.

Therefore the profile should be steeper as the amplitude of the sustained response increases. This conclusion was confirmed by the observation that the convergence of the response profiles became steeper with increasing sustained depolarization (Fig. 5B). Furthermore, the amplitude of each EPSP with each stimulus decreased as the sustained depolarization developed (Fig. 5C). The -burst stimulation might reduce extracellular K⁺ build-up, eliminating the conduction block in the presynaptic fiber (Fig. 5D) and the blockage of electrical conduction in pyramidal cells (Fig. 5E), resulting in pairing of somatic spikes and dendritic depolarization. These data support the idea of a sustained depolarization-dependent conduction block in presynaptic fibers. Of course, there are other possibilities to account for the steep convergence of the response along the transverse axis, such as an accumulation of the activity of the recurrent neural networks. However, this might counteract the depolarizing GABA_A receptor potential and accumulation of extracellular K⁺.

Physiological implications

Awareness has increased recently about the implications of time-encoded neural bursts (including the -burst in the hippocampus) in learning and memory (Fox 2000; Paulsen and Sejnowski 2000). The difference in the sustained depolarization elicited by standard or -burst stimulation could offer a physiological explanation for the different types of spike bursts and their roles in computation in the brain.

Tetanic stimulation caused spatially unequal sustained depolarization in area CA1 (Fig. 5, A and B). This inequality could explain the locus dependence of intracellular signals (Autere et al. 1999) and molecular mechanisms, such as the locus-dependent molecular mechanism of NMDA-independent LTP (Grover 1998).

Our results also have interesting information processing implications for hippocampal learning and memory function. Because the absence of CA1 pyramidal cell output was observed after standard tetanic stimulation, our results suggest that area CA1 behaves like a temporal filter for the inter-regional transmission of 100-Hz stimuli (Fig. 5, A, B, and E), which is physiologically implemented via a conduction block of presynaptic fibers or other mechanisms (Fig. 5C). In other words, neural information encoded in a 100-Hz signal transmitted from area CA3 to CA1 is severely transformed in CA1 before being transmitted to other areas because only the first part of a tetanus is faithfully transmitted (Fig. 3). At the same time, however, this type of signal facilitates subsequent information flow between the two areas through synaptic strengthening, evidenced by LTP. These physiological consequences of certain stimulation frequencies might highlight an artificial nature of the tetanic stimulation used experimentally but also could suggest a new functional role of the hippocampus in processing time-encoded neural information (Fox 2000).