The Journal of Neurophysiology Vol. 77 No. 5 May 1997, pp. 2466-2483
Copyright ©1997 by the American Physiological Society

Local and Propagated Dendritic Action Potentials Evoked by Glutamate Iontophoresis on Rat Neocortical Pyramidal Neurons

Peter C. Schwindt and Wayne E. Crill

Department of Physiology and Biophysics, University of Washington School of Medicine, Seattle, Washington 98195-7290

	ABSTRACT

Abstract Introduction Methods Results Discussion References

Schwindt, Peter C. and Wayne E. Crill. Local and propagated dendritic action potentials evoked by glutamate iontophoresis on rat neocortical pyramidal neurons. J. Neurophysiol. 77: 2466-2483, 1997. Iontophoresis of glutamate at sites on the apical dendrite 278-555 µm from the somata of rat neocortical pyramidal neurons evoked low-threshold, small, slow spikes and/or large, fast spikes in 71% of recorded cells. The amplitude of the small, slow spikes recorded at the soma averaged 9.1 mV, and their apparent threshold was <10 mV positive to resting potential. Both their amplitude and their apparent threshold decreased as the iontophoretic site was moved farther from the soma. These spikes were not abolished by somatic hyperpolarization. When the somata of cells displaying these small spikes were voltage clamped at membrane potentials that prevented somatic or axonic firing, corresponding current spikes could be evoked all-or-none by dendritic depolarization, indicating that the small, slow spikes arose in the dendrite. Similar responses were not observed during somatic depolarization evoked by current pulses or glutamate iontophoresis. These small, slow spikes were abolished by blocking voltage-gated Ca²⁺ channels but not by blocking Na⁺ channels or N-methyl-D-aspartate receptors. We conclude that these Ca²⁺ spikes occurred in a spatially restricted region of the dendrite and were not actively propagated to the soma. In the presence of 10 mM tetraethylammonium chloride, the amplitudes of the iontophoretically evoked Ca²⁺ spikes were large, similar to those of the Ca²⁺ spikes evoked by somatic current injection, but their apparent thresholds were 63% lower. We conclude that dendritic K⁺ channels normally prevent the active propagation of Ca²⁺ spikes along the dendrite. In 36% of recorded cells dendritic glutamate iontophoresis evoked a Na⁺ spike with an apparent threshold 63% lower than those evoked by somatic current injection or somatic glutamate iontophoresis. Blockade of these low-threshold Na⁺ spikes by pharmacological or electrophysiological means often revealed underlying small dendritic Ca²⁺ spikes. When cells displaying the low-threshold Na⁺ spikes were voltage clamped at membrane potentials that prevented firing of the soma or axon, corresponding tetrodotoxin-sensitive current spikes could be evoked all-or-none by dendritic depolarization. We conclude that these low-threshold Na⁺ spikes were initiated in the dendrite, probably by local Ca²⁺ spikes, and subsequently propagated actively to the soma. Most cells displaying dendritic Na⁺ spikes fired multiple bursts of action potentials during tonic dendritic depolarization, whereas somatic depolarization of the same cells evoked only regular firing. We discuss the implications of dendritic Ca²⁺ and Na⁺ spikes for synaptic integration and neural input-output relations.

	INTRODUCTION

Abstract Introduction Methods Results Discussion References

Spencer and Kandel (1961) proposed that a local, nonpropagated Na⁺-dependent action potential was generated at a site in the dendrite and served to provide an all-or-none "boost" to sufficiently large excitatory postsynaptic potentials (EPSPs) arriving at this site. This idea has remained popular because much evidence for the occurrence of dendritic spikes has accumulated, but ideas about the site of dendritic spike initiation and the nature of the dendritic action potential have changed. Intradendritic recordings from cerebellar Purkinje cells suggested that many dendritic sites were capable of initiating spatially restricted (local) Ca²⁺-dependent action potentials (Llinas and Sugimori 1980), and this has been supported by subsequent Ca²⁺ imaging studies (Ross et al. 1990). In the Purkinje cells, the local dendritic Ca²⁺ spike acted like a "giant EPSP" that triggered a burst of Na⁺-dependent action potentials at the soma (Llinas and Sugimori 1980). These observations are consistent with the boosting idea, because synaptic input can initiate the localized Ca²⁺ spikes (Miyakawa et al. 1992), but they also are consistent with many dendritic boosting sites instead of one site. Similarly, intrinsic bursts of Na⁺-dependent spikes in hippocampal CA3 neurons are thought to be driven by a dendritic Ca²⁺-dependent depolarization (Wong and Prince 1978), but dendritic recordings (Benardo et al. 1982; Magee and Johnston 1995a,b; Wong and Stewart 1992) and imaging studies (Jaffe et al. 1992; Magee et al. 1995; Regehr et al. 1989) have suggested that Ca²⁺ influx through voltage-gated channels can occur over a wide area of the hippocampal neuron dendrite rather than at a few discrete sites.

Ca²⁺ and Na⁺ spikes have been recorded in the dendrites of neocortical pyramidal neurons (Amitai et al. 1993; Kim and Connors 1993; Pockberger 1991; Stuart and Sakmann 1994). Ca²⁺ imaging studies also have revealed that the whole dendritic tree of neocortical pyramidal neurons contains voltage-gated Ca²⁺ channels (Markram and Sakmann 1994; Markram et al. 1995; Schiller et al. 1995; Yuste et al. 1994). Imaging studies also have suggested that much or all of the dendritic membrane of pyramidal neurons contains a sufficient density of Na⁺ channels to support propagated Na⁺ spikes (Jaffe et al. 1992; Markram et al. 1995; Schiller et al. 1995). Because Ca²⁺ and Na⁺ channels are present over most or all of the dendritic tree, it seems possible that Ca²⁺ or Na⁺ spikes might arise anywhere on the dendritic tree where the channels were activated by adequate depolarization. When evoked by brief intracellular injected current pulses, the Na⁺ spike appears to originate distal to the soma and invades the dendrite subsequently (Stuart and Sakmann 1994), but other studies suggest that adequate orthodromic input can initiate Na⁺ spikes in the dendrite that then propagate to the soma (Colling and Wheal 1994; Poolos and Kocis 1990; Regehr et al. 1993; Turner et al. 1991; Wong and Stewart 1992).

The question of dendritic spike generation is important because it influences our ideas about how a neuron processes synaptic input, and it determines how well we can study a single neuron's input-output relation with the use of standard intracellular techniques. In the classical view derived from studies of cat lumbar motoneurons (Eccles 1957), the neuron is integrative. The summation of many individual EPSPs is needed to initiate an action potential, which arises at a site downstream from all synaptic input, and the effectiveness of the EPSPs is weighted according to their electrotonic distance from this site. Because the downstream spike initiation site cannot tell the difference between current injected from a microelectrode and synaptic current arriving from the dendrites, we can determine the neuron's input-output relation by somatic current injection. In contrast, the boosting idea implies at least two sites for spike initiation, or perhaps an infinite number of sites according to modern ideas of Ca²⁺ and Na⁺ channel distribution. Each site may have a different input-output relation, which cannot be discovered by somatic current injection if the sites are remote from the soma (Wong and Stewart 1992). If excitatory synaptic input is only boosted locally, the downstream summing site may still initiate the spikes that are propagated down the axon, and there is some semblance of integrative behavior. If propagated spikes can arise anywhere on the dendritic tree, it is as if the neuron were one large isopotential cell. Only the local threshold at a dendritic site would be important, and when a local synaptic input was large enough a spike would travel down the axon, no matter where the input occurred.

With the use of a long-lasting iontophoresis of glutamate on the apical dendrite of rat neocortical pyramidal cells, we found that the steady-state axial current transmitted to the soma during dendritic depolarization was amplified by dendritic voltage-gated ion channels (Schwindt and Crill 1995), and we have shown that this amplification of tonic transmitted current is physiologically relevant to tonic repetitive firing (Schwindt and Crill 1996). During these experiments we also observed transient spike responses, which we describe here. The nature of these spike responses raises the possibility that each of the three models of neural information processing outlined above holds under different conditions in neocortical pyramidal cells.

	METHODS

Abstract Introduction Methods Results Discussion References

Methods were similar to those described previously (Schwindt and Crill 1995, 1996). Sprague-Dawley rats of either sex (21-35 days postnatal) were anesthetized with ketamine (150 mg/kg) and xylazine (10 mg/kg) and killed by carotid section. A coronal section of cortex 0-3 mm posterior to bregma was isolated, and slices 350-µm thick were prepared and maintained as described. Recorded cells lay 0.89-1.30 mm below the pial surface (mode: 1.18 mm) and 2.04-3.18 mm from midline (mode: 2.78 mm), corresponding to layer 5 of areas HL and FL of sensorimotor cortex (Zilles and Wree 1985). Three cells in this study were recovered after being injected with biocytin (0.5% in 2.7 M KCl) and were visualized after histological processing. Similar to results from a larger population of stained cells (Schwindt and Crill 1995), each had a pyramidal-shaped soma in deep layer 5 and an apical dendrite extending to the pial surface with a terminal tuft and with the first major branch point 500-600 µm from the soma (see Fig. 1A).

View larger version (25K): [in this window] [in a new window]   FIG. 1. Properties of recorded cells. A: camera lucida drawing of a portion of a biocytin-stained neuron in which much of the apical dendrite was contained in a single histological section. Main features are typical of recovered biocytin-stained cells. Arrow: site of glutamate iontophoresis in this neuron on the basis of measurements of electrode position during recording, which was 1 of the most distal sites employed. Most iontophoretic sites in this study were closer to the soma. B: subthreshold voltage responses (top) to constant injected current pulses (bottom) show typical features observed in each cell. C: plot of membrane potential vs. injected current for cell in B. Plots for peak voltage deflection () and voltage at 1 s () are shown and fit with 2 lines whose slopes give input resistance for each response and range of membrane potential.

Recordings were made in a chamber with the slice submerged and maintained at 33-34°C. Slices were perfused with a physiological saline consisting of (in mM) 130 NaCl, 3 KCl, 2 CaCl₂, 2 MgCl₂, 1.25 NaH₂PO₄, and 10 dextrose saturated with 95% O₂-5% CO₂, pH 7.4. Tetrodotoxin (TTX, 1 µM), D2-amino-5-phosphonopentoic acid (APV, 100 µM) or MK801 (10 µM) were added to this perfusate. Tetraethylammonium chloride (TEA, 10 mM) was substituted for 10 mM NaCl. To block voltage-gated Ca²⁺ channels, one of the following ion substitutions was employed in a particular experiment: 1 or 2 mM MnCl₂, 2 mM NiCl₂, or 200 µM CdCl₂ was substituted for CaCl₂ in an equimolar amount and NaH₂PO₄ was omitted to avoid precipitation.

Cells were impaled with sharp microelectrodes made from standard borosilicate tubing (1.0 mm OD). In most experiments they contained 2.7 M KCl (DC resistance 30-40 M). During iontophoresis, somatic membrane potential was maintained constant with the use of an Axoclamp-2A amplifier (Axon Instruments, Foster City, CA) in single-electrode voltage-clamp mode with a switching rate of 2.5-5.5 kHz (30% duty cycle). This voltage-clamp amplifier allows the recording of the actual membrane potential maintained during voltage clamp, and series resistance is eliminated inherently by its mode of operation.

The second current-clamp amplifier and headstage of the same Axoclamp-2A was used to pass a constant current through the iontophoretic microelectrode. Iontophoretic microelectrodes were broken to a tip diameter of ~2 µm and filled with 0.5 M sodium glutamate buffered to pH 7.4 with 30 mM N-2-hydroxyethyl piperazine-N-2-ethanesulfonic acid. Negative iontophoretic currents of 15-100 nA (mode: 50 nA) were employed from a +5-nA holding current. Positive iontophoretic currents were tested but never produced a postsynaptic response. The iontophoretic electrode was positioned with the use of a separate micromanipulator, and an effective site was found near a line extending from the recording electrode normal to the pial surface. Vertical electrode movements of ~10 µm caused response amplitude to vary from zero to maximum. Postsynaptic responses, evoked by an iontophoresis 100 ms-2 s in duration repeated each 15-20 s were stable and reproducible (cf. Hu and Hvalby 1992). The distance between recording and iontophoretic electrodes was measured at the slice surface with the use of a calibrated eyepiece on a dissecting microscope.

Membrane potential, membrane current, and iontophoretic current were monitored, amplified, and recorded on a multichannel video cassette recorder with pulse code modulation (Neuro-Data, New York, NY). Membrane potential and injected current were filtered at 2-10 kHz. Membrane current evoked during voltage clamp was filtered at 100 Hz. Resting potential was taken as the difference between the intracellular and extracellular potentials recorded on a chart recorder. In some experiments the time derivative (dV/dt) of recorded membrane potential was computed by leading membrane potential into an electronic differentiator whose voltage output was proportional to dV/dt. Recorded data were played back into a storage oscilloscope for photography or digitized for further analysis by computer.

	RESULTS

Abstract Introduction Methods Results Discussion References

Cell properties

Cells were accepted for analysis only if they exhibited stationary resting potentials and responses to iontophoresis both during the control period and after the application of any pharmacological agents. Data were obtained from 57 cells meeting these criteria. Resting potential ranged from 67 to 80 mV (mode: 70 mV) and was little affected by the pharmacological agents employed in this study (see METHODS). All cells displayed a relaxation of membrane potential back toward resting potential during the application of a 1-s hyperpolarizing current pulse, but rarely during depolarization (Fig. 1B). Plots of membrane potential versus injected current were well fit by two lines intersecting near resting potential (Fig. 1C). The slopes of these lines gives input resistance, which at the end of the 1-s pulse averaged 27.6 M for depolarization (range: 13.2-90.1 M) and 15.8 M for hyperpolarization (range: 6.5-38.3 M). On average, steady-state input resistance during depolarization was 1.7 times greater than during hyperpolarization. A sufficiently large injected current pulse 1 s in duration evoked tonic repetitive firing in all cells. Thirty-five percent of the cells displayed an initial burst of action potentials at the onset of the pulse.

Dendritic spikes evoked by iontophoresis

When glutamate was iontophoresed on the apical dendrite at a site between 278 and 555 µm from the soma, unusual early responses were observed during recordings from 41 of 57 cells. These unusual responses consisted of membrane potential oscillations or low-threshold spikes, or a combination of the two, which could not be duplicated by injection of constant current pulses into the soma of the same cell. Figure 2 illustrates some features of the early membrane potential oscillations. In Fig. 2A the strength of a 200-ms iontophoresis applied 463 µm from the soma was adjusted to evoke a small abrupt depolarization in an all-or-none manner. The dendritic iontophoresis was followed in the same sweep by a current ramp injected into the soma that was just large enough to evoke a spike. The abrupt iontophoretically evoked depolarization triggered a large, fast spike whose firing level was similar to the spike evoked by intrasomatic current injection. By firing level we mean the somatic membrane potential just before the rapid upstroke of the Na⁺ spike, and we take this as a measure of spike threshold at that instant in time. On about every other sweep the iontophoresis evoked an abrupt depolarization that triggered two spikes (Fig. 2B). In Fig. 2C, the soma was hyperpolarized ~15 mV by injection of direct current, and a slow, small oscillation (*) was revealed. This oscillation was evoked in an all-or-none manner, indicating that it represented an action potential. The apparent threshold of this small, slow spike (the membrane potential at which "all" and "none" responses diverge) is negative to resting potential and far negative to firing level of the current-evoked spike. The spike evoked by intrasomatic current injection is presumed to arise in the axon initial segment (Eccles 1957) or farther down the axon (Colbert and Johnston 1996). Because the small, slow spike was evoked when the soma was hyperpolarized, and because it was evoked only by depolarization of the dendrite, it seems likely that it represents an action potential arising in the dendrite.

View larger version (15K): [in this window] [in a new window]   FIG. 2. Small, slow spike and large, fast, low-threshold spikes evoked by depolarization of apical dendrite but not soma of same cell. A: oscilloscope records of membrane potential responses (top trace) to glutamate iontophoresis (bottom trace) on apical dendrite 463 µm from soma followed by intrasomatic current injection (middle trace) during same sweep. Iontophoretic strength was adjusted to evoke small abrupt depolarization all-or-none on consecutive sweeps. Fast spikes are truncated in all records. B: during other sweeps the dendritic depolarization evoked 2 spikes. Bottom trace: low-gain time derivative (dV/dt) of membrane potential, which marks spike initiation and shows all spikes had similar maximum rates of rise. C: hyperpolarization by injected direct current (bottom trace) eliminated fast spikes evoked by dendritic iontophoresis and revealed small, slow spike. D: glutamate iontophoresis on soma of same cell followed by injected current. Iontophoretic and injected currents were adjusted to evoke spikes all-or-none. Vertical calibration bar: 20 mV, 1 nA for injected current; 200 nA for iontophoretic current; 1,000 V/s for dV/dt. Horizontal bar: 200 ms for A, B, and D; 100 ms for C.

In this cell, we also iontophoresed glutamate at a site ~20 µm from the recording electrode and at a similar depth below the slice surface, i.e., on or near the soma. No oscillations or low-threshold spikes were evoked by iontophoresis at this site, and the iontophoretically evoked and current-evoked spikes had similar firing levels (Fig. 2D). We compared the firing level of spikes evoked by somatic current pulses and by glutamate iontophoresis on the somata (only) of four other cells. In these experiments the iontophoretic electrode also was inserted within 20 µm of the recording electrode, but APV (100 µM) was present to block N-methyl-D-aspartate (NMDA) receptors. The presence of APV allowed a more critical test of whether we were iontophoresing on the soma. Unlike penetrations at distal locations from the soma, we found many responsive sites when the iontophoretic electrode was placed in the vicinity of the soma, undoubtedly caused by the stimulation of the many basal and proximal apical oblique dendrites that surround the soma in a thick slice. In these experiments a somatic site was recognized by the linear decrease of the glutamate-evoked current with somatic depolarization instead of the voltage-dependent amplification of transmitted current observed during dendritic iontophoresis (Schwindt and Crill 1995). The APV ensured that the nonlinear increase of glutamate current with depolarization caused by stimulation of NMDA receptors (Flatman et al. 1986; Schwindt and Crill 1995) would not confound the recognition of a somatic site. Neither membrane potential oscillations nor low-threshold spikes were observed during somatic iontophoresis, and the firing levels of spikes evoked by injected current and by somatic glutamate iontophoresis were identical (data not shown).

Long-lasting glutamate iontophoresis on the apical dendrite usually evoked a series of early oscillations, as illustrated for one cell in Fig. 3. In most cells in which these early oscillations were observed, firing level was not reached. In Fig. 3A are superimposed the response to a current pulse injected into the soma (trace 1) and the response to the dendritic iontophoresis 370 µm from the soma (trace 2, shown at 16 times slower sweep speed). During the dendritic depolarization, somatic membrane potential exhibited a series of initial oscillations (Fig. 3A, asterisks) before reaching a final steady level. This is a typical feature. The oscillations were observed only during the early portion of a long iontophoresis in all recorded cells. They never were observed when the transmitted current was steady, a period lasting up to 2 s in some cells. These observations suggest that the mechanism underlying the oscillations is activated transiently and normally inactivates during steady dendritic depolarization.

View larger version (20K): [in this window] [in a new window]   FIG. 3. Early membrane potential oscillations during long-lasting dendritic depolarization associated with all-or-none current spikes during voltage clamp below somatic firing level. A: superimposed records of membrane potential (top traces) evoked by 1 s of glutamate iontophoresis (trace 2) on apical dendrite 370 µm from soma and repetitive firing (trace 1) evoked by intrasomatic current injection. Repetitive spikes are truncated. Asterisks: initial membrane potential oscillations evoked by dendritic depolarization. B: superimposed records of same iontophoretically evoked response as in A and membrane potential recorded when iontophoresis was repeated during DC voltage clamp of soma at indicated potentials. C: membrane current evoked by iontophoresis during voltage clamp at somatic holding potentials shown in B. Asterisks: early, transient, current spikes observed during iontophoresis when soma was voltage clamped at 68 mV. Horizontal dashed line: DC baseline current at 68 mV before iontophoresis. I: deviation from this baseline caused by axial current from dendrite arriving at soma. Vertical scale and time base of records in B also apply to A, except that repetitive firing trace in A is 16 times faster.

The superposition of current-evoked and the iontophoretically evoked responses in Fig. 3A shows that the iontophoretically evoked oscillations did not reach the firing level of spikes evoked by intrasomatic current injection. Somatic membrane potential was subsequently held at different steady levels by voltage clamp, the glutamate iontophoresis was repeated, and the axial current transmitted from dendrite to soma during the iontophoresis was measured. Figure 3B shows a record of the same iontophoretically evoked response as in Fig. 3A, on which are superimposed records of the somatic membrane potential during voltage clamp(at 68 and 74 mV) when the iontophoresis was repeated. The holding potentials shown are not simply command potentials. Because we used discontinuous single-electrode voltage clamp (see METHODS), the actual membrane potential attained during voltage clamp could be measured. Figure 3C shows the current measured when the iontophoresis was repeated at the two holding potentials in Fig. 3B. The steady baseline current at a given holding potential (indicated by dashed line in Fig. 3C for a holding potential of 68 mV) is generated in part by any noninactivating, voltage-gated channels in the soma or axon that may have been activated at the holding potential. The depolarizing current transmitted from dendrite to soma during the iontophoresis is given by the downward deflection of the current from this baseline (indicated by I in Fig. 3C). Because of its cablelike structure, membrane potential along the apical dendrite will not be the same as the potential at the soma, but altering somatic membrane potential will nevertheless alter membrane potential along some portion of the apical dendrite (Rall and Segev 1985). Indeed, we have presented evidence that we can alter dendritic membrane potential at least out to 300-400 µm from the soma by altering somatic membrane potential (Schwindt and Crill 1995). In the cell of Fig. 3 we employed this indirect method of altering dendritic potential to show that the oscillations represent all-or-none action potentials, and, furthermore, that these action potentials can be evoked by dendritic depolarization at the same time action potentials are prevented from arising in the soma or axon.

When somatic membrane potential was held negativeto 68 mV (e.g., at 74 mV in Fig. 3B) and the dendritic iontophoresis was applied, the transmitted current waveform was smooth. When membrane potential was depolarized to a level near where the oscillations arose in the current-clamp recording (at 68 mV in Fig. 3B) and the iontophoresis was repeated, "current spikes" (downward deflections marked by asterisks in Fig. 3B) were observed on the current recording. That is, we evoked the current spikes in an all-or-none manner by indirectly manipulating dendritic potential. This shows that the current spikes recorded in voltage clamp represent all-or-none action currents underlying action potentials. The current spikes appeared only when the dendrite was depolarized by the iontophoresis: depolarizing the soma to 68 mV in the absence of dendritic depolarization did not evoke the current spikes, only the steady baseline current. Because somatic membrane potential during the entire iontophoretic response was below firing level (Fig. 3A), it is difficult to see how these spikes could arise from the soma or axon. Clamping somatic membrane potential negative to firing level (Fig. 3B) prevented the initiation of a spike in the soma or axon in any case. Thus the spikes evoked by the dendritic depolarization must have a dendritic origin. The most likely initiation site of these dendritic spikes is the region where the dendrite is most depolarized, namely, at or near the site of iontophoresis.

Spatially restricted dendritic Ca²⁺ spikes

Figure 4, A-C, shows an example of a small, slow spike evoked by dendritic depolarization in another cell. Iontophoretic strength was adjusted to evoke the spike an all-or-none manner (Fig. 4A). The spike was abolished by substitution of 1 mM Mn²⁺ for 1 mM Ca²⁺, even when iontophoretic strength (and somatic depolarization) was increased above the control value (Fig. 4B). In this cell, the Mn²⁺ was washed out by normal physiological saline containing 1 µM TTX, and the spike reappeared during a low-strength iontophoresis (Fig. 4C). Similar results were obtained in each of six cells tested in this way. In eight other cells in which a divalent cation (Mn²⁺, Ni²⁺, or Cd²⁺) was partially or fully substituted for Ca²⁺ (see METHODS), the current spikes observed during voltage clamp (as in Fig. 3, B and C) also were abolished reversibly (data not shown). In two cells we also ascertained that the iontophoretically evoked spike survived application of 100 µM APV or 10 µM MK801. This test of whether the spikes could survive blockade of NMDA receptors was prompted by the fact that the inward current flowing through NMDA channels, which are voltage dependent in the presence of extracellular Mg²⁺, can also result in a regenerative response (Flatman et al. 1986). When NMDA receptors were blocked, a higher iontophoretic strength was required to evoke the spike, but its voltage threshold was identical to control (data not shown). Thus these iontophoretically evoked dendritic spikes are Ca²⁺ dependent.

View larger version (15K): [in this window] [in a new window]   FIG. 4. Small Ca2+ spike evoked by dendritic depolarization. A: digitized records of small spike potential (top trace) evoked all-or-none by a short glutamate iontophoresis (bottom trace) on apical dendrite 407 µm from soma. B: spike was abolished when Mn2+ was substituted for Ca2+ in the perfusate even when a stronger iontophoresis was applied to cause greater depolarization. C: spike reappeared when Ca2+-containing control solution, which also contained 1 µM tetrodotoxin (TTX), was reapplied. D: plot of Ca2+ spike amplitude, measured from threshold to peak, evoked by dendritic glutamate iontophoresis at various distances from soma in 22 cells. Key indicates those Ca2+ spikes measured in physiological saline, those measured in saline containing 1 µM TTX, and those that were tested and subsequently abolished by partial or full substitution of Ca2+ by another divalent cation as explained in METHODS.

Interesting aspects of these Ca²⁺ spikes were their small size and low apparent threshold, which was <10 mV positive to resting potential. These Ca²⁺ spikes were much smaller than those typically evoked by somatic depolarization in TEA-containing saline (see Fig. 7). Their mean amplitude, measured from threshold to peak, was 9.1 ± 4.3 (SD) mV, but their amplitude depended partly on the site of iontophoresis, as shown in the plot of Fig. 4D. Ca²⁺ spike amplitude decreased with iontophoretic distance out to 300-400 µm from the soma, after which a relation to iontophoretic distance was no longer clear. This plot also indicates that Ca²⁺ spike amplitude was about the same whether measured in TTX or physiological saline. The duration of Ca²⁺ spikes (measured at threshold) averaged 92 ± 47 (SD) ms and showed no relation to iontophoretic distance (data not shown).

View larger version (8K): [in this window] [in a new window]   FIG. 7. Large Ca2+ spike evoked by dendritic depolarization in presence of 10 mM tetraethylammonium chloride (TEA) and 1 µM TTX. All records from same cell. A: superimposed oscilloscope records of large Ca2+ spike evoked all-or-none (top trace) by glutamate iontophoresis (bottom trace) on apical dendrite 307 µm from soma. Arrow: apparent threshold of Ca2+ spike. B: iontophoresis was preceded by somatic current injection (middle trace) that evoked a depolarization just subthreshold for somatic initiation of Ca2+ spike. C: large Ca2+ spike evoked all-or-none by current injection in soma. Arrow: somatic Ca2+ spike threshold. Note faster time base in C compared with A and B.

In one cell we were able to measure the responses to iontophoresis at different distances from the soma in physiological saline containing 1 µM TTX (Fig. 5). Figure 5, A1, B1, and C1, shows membrane potential responses to iontophoresis at the indicated distances from the soma. Figure 5, A2, B2, and C2, shows corresponding records of the transmitted current measured during somatic voltage clamp at various holding potentials. At each iontophoretic distance we were able to evoke early spikes (and a late plateau depolarization) in an all-or-none manner by adjusting the iontophoretic strength (e.g., Fig. 5, A1 and C1). The corresponding voltage-clamp records were taken with the use of the iontophoretic strength that resulted in the spikes and plateau. Spike (and plateau) amplitude decreased as the iontophoretic site was moved farther from the soma (Fig. 5, A1, B1, and C1; note different voltage calibrations in each panel). At the closest site, no current spikes were seen during voltage clamp (Fig. 5A2), suggesting that membrane potential was controlled well enough between the soma and the iontophoretic site to prevent uncontrolled spike activity. At the two more distal sites uncontrolled current spikes were seen during voltage clamp (Fig. 5, B2 and C2), and current spike amplitude was smaller at the more distal site (note different current calibrations in each panel). Thus both voltage and current spikes decreased as the iontophoretic site was moved farther from the soma, and the prevention of spike initiation by voltage clamp also was lost as the site moved distally. By inspecting the records in Fig. 5, A1, B1, and C1, it can also be seen that the apparent threshold of the spikes above resting potential (where the all and none responses diverge) also decreased with intophoretic distance.

View larger version (39K): [in this window] [in a new window]   FIG. 5. Ca2+ spike amplitude and apparent threshold decreased as site of dendritic depolarization was moved farther from soma in a single cell. All records from same cell in solution containing 1 µM TTX. A1, B1, and C1: superimposed digitized records of membrane potential responses (top traces) to 2 iontophoretic strengths (bottom traces) that evoked early membrane potential spikes (and a late plateau potential) in an all-or-none manner. Records in A1, B1, and C1 were obtained during iontophoresis on apical dendrite at indicated distances from soma. Note difference voltage calibration for each panel. A2, B2, and C2: superimposed records of membrane currents (top traces) recorded in voltage clamp during the larger iontophoresis indicated in A1, B1, and C1, respectively, over a range of somatic holding potentials ( at bottom). Note different current calibrations for each panel.

It is difficult to imagine how a spike of a few millivolts in amplitude could actively propagate to the soma. A more reasonable interpretation of these observations is that the Ca²⁺ spikes are generated in a restricted area of the dendrite. Their small amplitude and low threshold when viewed from the soma is likely the result of electrotonic decrement, which attenuates a voltage transient generated at a distal site on a cablelike structure (Rall 1977), i.e., they appear small when viewed from the soma because they are not actively propagated to the soma. At their site of origin, the spikes may be much larger and their true threshold much higher.

Dendritic Ca²⁺ spikes are influenced by somatic Na⁺ spikes

Figure 6 shows an interesting property of the local dendritic Ca²⁺ spike that was observed in each of three cells tested. In Fig. 6A a small spike was evoked all-or-none by adjusting iontophoretic strength during a short iontophoresis. Figure 6B illustrates our finding that the dendritic spike failed if it was preceded by a Na⁺ spike evoked at the soma by an injected current pulse within a certain time interval. When performing this test, iontophoretic strength was increased to result in a dendritic spike on every trial if not preceded by a somatically evoked spike. At this time interval, the dendritic spike was present when the somatic depolarization was subthreshold for spike initiation (trace 2) but was absent when the somatic spike was evoked (trace 1). Interestingly, if the interval between the two spikes was lengthened, the dendritic spike was larger when it was preceded by the somatic spike (Fig. 6C). These results indicate that dendritic spike initiation can be suppressed or enhanced by somatic spike initiation. This interaction may have functional significance during glutamate-driven repetitive firing when both dendritic and somatic spikes are being initiated (see DISCUSSION).

View larger version (9K): [in this window] [in a new window]   FIG. 6. Dendritic Ca2+ spike influenced by somatic Na+ spike. All records from same cell. A: superimposed oscilloscope records of small spike evoked all-or-none (top trace) by brief glutamate iontophoresis (bottom trace) on apical dendrite 315 µm from soma. B and C: iontophoretic strength was increased (bottom traces) to consistently evoke a small spike, and small spike was preceded by brief current pulses injected into the soma (middle traces) that either evoked a Na+ spike (trace 1) or were just subthreshold for Na+ spike initiation (trace 2). Records corresponding to suprathreshold and subthreshold injected current pulses are superimposed in B and C, and evoked Na+ spikes are truncated. B: small spike was occluded when somatic current injection evoked an action potential (trace 1), but reappeared if the somatic depolarization was just subthreshold (trace 2). C: when interval between somatic and dendritic spikes was lengthened, the somatic action potential was followed by a dendritic spike (trace 1) that was larger than the spike that followed a just subthreshold somatic depolarization (trace 2).

Dendritic K⁺ channels prevent active propagation of dendritic spikes

There is evidence from Ca²⁺ imaging studies that voltage-gated Ca²⁺ channels exist along the entire dendritic membrane (e.g., Schiller et al. 1995; Yuste et al. 1994). What then would prevent the active propagation of a dendritic Ca²⁺ spike once it was initiated? Because we know that large Ca²⁺ spikes can be evoked by somatic depolarization in neocortical pyramidal neurons after K⁺ currents are reduced by TEA application (Reuveni et al. 1993; Stafstrom et al. 1985), we hypothesized that dendritic K⁺ channels normally prevent the active propagation of the Ca²⁺ spikes along the dendrite. This idea was tested by depolarizing the apical dendrite in the presence of 10 mM TEA and 1 µM TTX. Typical results obtained in each of five cells tested are shown in Fig. 7. In the presence of TEA and TTX, iontophoretic strength always could be adjusted to evoke a large Ca²⁺ spike in an all-or-none manner (Fig. 7A). The amplitude of the iontophoretically evoked Ca²⁺ spikes in TEA (measured from resting potential to peak) averaged 86 mV (range: 76-90 mV), and was within a few millivolts of the amplitude of Ca²⁺ spikes evoked by depolarization of the soma in the same cell (cf. Fig. 7C). These large spikes are identified as Ca²⁺ spikes because TTX was present to block Na⁺ currents, and the spikes were abolished by substitution of Mn²⁺ for Ca²⁺ in the perfusate (data not shown). In addition, NMDA receptor blockers were present in three of the experiments (100 µM APV, n = 2; 10 µM MK801, n = 1) to ensure that current flowing through voltage-dependent NMDA channels did not contribute to the regenerative responses.

Although the iontophoretically evoked Ca²⁺ spikes had a large amplitude in TEA, their apparent threshold was still quite low. For example, the first depolarization in Fig. 7B was evoked by injecting a current pulse in the soma that was just subthreshold for the somatically evoked Ca²⁺ spike (cf. Fig. 7C). This subthreshold depolarization is much larger than the apparent threshold of the Ca²⁺ spike evoked by iontophoresis (Fig. 7A, arrow), which is taken at the membrane potential at which the all and none responses diverge. For the five cells examined, the apparent depolarization from resting potential at which the iontophoretically evoked Ca²⁺ spikes arose averaged 36% of the somatic depolarization needed to evoke a Ca²⁺ spike by somatic current injection in the same cell. The similar amplitudes but differing thresholds of the current-evoked and iontophoretically evoked Ca²⁺ spikes are consistent with the idea that the iontophoretically evoked spikes were initiated in the dendrite and, because dendritic K⁺ currents were reduced by TEA, propagated actively to the soma.

Dendritic Na⁺ spikes

In addition to the Ca²⁺ spikes described above, we observed low-threshold Na⁺ spikes evoked by dendritic depolarization in 21 cells examined in current clamp or voltage clamp. An example of a low-threshold Na⁺ spike is shown in Fig. 8. A dendritic glutamate iontophoresis 200 ms in duration evoked a low-threshold Na⁺ spike (Fig. 8A, arrow) that was evoked all-or-none by adjusting iontophoretic strength. The apparent firing level of this spike was far below that of the spike evoked by somatic current injection (Fig. 8A, right) but had a similar amplitude and duration (Fig. 8, B1 and B2). When we iontophoresed glutamate at a site ~20 µm from the recording electrode and at a similar depth below the slice surface (i.e., on or near the soma), no low-threshold spikes were evoked, and the iontophoretically evoked and current-evoked spikes had similar firing levels (Fig. 8C).

View larger version (7K): [in this window] [in a new window]   FIG. 8. Low-threshold Na+ spike evoked by depolarization of dendrite but not soma of same cell. A: oscilloscope records of membrane potential responses (top trace) to glutamate iontophoresis (bottom trace) on apical dendrite 300 µm from soma followed by intrasomatic current injection (middle trace) during same sweep. Arrow: iontophoretically evoked low-threshold spike. Iontophoretic and injected currents were adjusted to evoke both spikes all-or-none on consecutive sweeps. Spikes are truncated in A and C. B: fast-sweep records of glutamate-evoked low-threshold spike (B1) and current-evoked spike (B2) whose peak is slightly clipped. C: glutamate iontophoresis on soma of same cell evoked spikes with same threshold as current-evoked spikes. Iontophoretic and injected currents were adjusted so both spikes were evoked all-or-none. Calibrations: horizontal bar in C, 200 ms for A and C, 2 ms for B; vertical bar in C, 20 mV for A and C, 40 mV for B, 1 nA for injected current, 200 nA for iontophoretic current.

The records of Fig. 9, from another cell, illustrate our finding that low-threshold Na⁺ spikes, like the Ca²⁺ spikes described above, were initiated only during the onset of a long-lasting dendritic depolarization. The first spike evoked during the long iontophoresis in Fig. 9A has a much lower firing level (arrow) than the subsequent spikes or the spikes evoked by somatic current injection in the same cell (Fig. 9B). The differences in spike firing levels are seen more clearly in the fast-sweep records of Fig. 9, C and D. The superimposed records of spikes in Fig. 9C are from the same record as Fig. 9A, but the oscilloscope trigger for each spike was set at the DC level corresponding to firing level (arrow) of the first spike evoked by the iontophoresis (labeled 1 in Fig. 9C). The same procedure was used for the current-evoked spikes of Fig. 9D, which are from the same record as Fig. 9B. The later spikes evoked by dendritic depolarization are grouped in Fig. 9C, right. Their firing level is similar to that of the current-evoked spikes of Fig. 9D but significantly more depolarized than the first iontophoretically evoked spike. The second and third spikes in Fig. 9C also have lower firing levels than the late spikes, and the second spike had the same firing level as the first during some applications of the same iontophoresis. In nine cells examined in current clamp that exhibited a low-threshold Na⁺ spike, the depolarization from resting potential at which the initial Na⁺ spike arose was only 36% as large as for the late spikes evoked by a long-lasting iontophoresis (7.4 vs. 20.1 mV positive to resting potential). In these nine cells, firing level of the late spikes was identical to firing level of spikes evoked by intrasomatic current injection, which are presumed to arise in the initial segment.

View larger version (27K): [in this window] [in a new window]   FIG. 9. Low-threshold Na+ spike evoked during onset of long-lasting dendritic depolarization. All (oscilloscope) records from same cell. A: membrane potential response (top trace) evoked by glutamate iontophoresis (bottom trace) on apical dendrite 444 µm from soma. Middle trace: low-gain dV/dt of membrane potential, which marks spike initiation and shows all spikes had similar maximum rates of rise. Arrow: firing level of 1st spike. B: record arranged similarly to A, but showing response to current pulse (bottom trace) injected into soma. C: superimposed records of spikes from sweep in A. Each sweep was triggered at DC level corresponding to firing level of 1st spike (arrow). First 3 spikes evoked by iontophoresis are labeled 1-3; spikes evoked later during iontophoresis are grouped at right. D: superimposed records of spikes from sweep in B triggered as in C. DC membrane potential levels are preserved in A-D. Calibrations: vertical bar in D, 20 mV, 1,000 V/s, 4 nA for injected current, 40 nA for iontophoretic current for A-D; horizontal bar in D, 200 ms for A, 100 ms for B, 10 ms for C and D.

Figure 10A shows that the low-threshold Na⁺ spike was not abolished by DC hyperpolarization of the soma. (It is evoked all-or-none by varying iontophoretic strength at the hyperpolarized potential.) These records are from the same cell as Fig. 9, but a single low-threshold spike was evoked by a short iontophoresis. Not only did this spike persist during somatic hyperpolarization, but its apparent threshold (indicated by  in Fig. 10A) was below resting potential. The membrane potential reached at the peak of the spike was not altered by the hyperpolarization (Fig. 10B). Both the spike's resistance to somatic hyperpolarization and its low apparent threshold are best explained if it was initiated at a site far from the soma. Because this spike was observed during depolarization of the dendrite but not during somatic depolarization, it seems reasonable to conclude that it was initiated in the dendrite and subsequently propagated actively to the soma.

View larger version (7K): [in this window] [in a new window]   FIG. 10. Low-threshold Na+ spike not blocked by somatic hyperpolarization. All (oscilloscope) records from cell of Fig. 9. A: membrane potential responses (top traces) to iontophoresis (bottom traces) at resting potential (trace 1) and when cell was hyperpolarized by injected direct current (trace 2). Iontophoretic current was decreased slightly during hyperpolarization to evoke low-threshold Na+ spike all-or-none. B: superimposed, fast-sweep records of like-labeled spikes of A. C: small spike evoked all-or-none (top traces) by dendritic iontophoresis (bottom trace) when cell was bathed in saline containing 1 µM TTX. Calibrations: horizontal bar in B, 100 ms for A and C, 2 ms for B; vertical bar in B, 10 mV for A, 40 mV for B, 5 mV for C, 40 nA for iontophoretic current in A and C.

In the cell of Figs. 9 and 10, a dendritic Ca²⁺ spike was not observed during iontophoresis in physiological saline. However, when 1 µM TTX was added to the saline and the iontophoresis was repeated, the low-threshold Na⁺ spike was abolished and replaced with a small spike (Fig. 10C) that was similar to the Ca²⁺ spikes described above. Thus it is possible that the dendritic Na⁺ spike was initiated by the spatially restricted dendritic Ca²⁺ spike. This idea was supported by observations in other cells, one of which is shown in Fig. 11. This figure also shows that somatic Na⁺ spikes can influence the initiation of dendritic spikes. Figure 11A shows that the iontophoretically evoked, low-threshold Na⁺ spike failed if it was preceded by a Na⁺ spike evoked at the soma within a certain time interval. When this test was performed, iontophoretic strength was adjusted to evoke a low-threshold Na⁺ spike on every trial if not preceded by a somatically evoked spike. At the time interval shown, the low-threshold Na⁺ spike was present when the somatic depolarization was just subthreshold (trace 2), but was absent when the somatic spike was evoked (trace 1). No sign of an underlying Ca²⁺ spike was seen when the low-threshold Na⁺ spike failed, but when the interval between stimuli was lengthened (Fig. 11B) the low-threshold Na⁺ spike failed and a small, presumed Ca²⁺ spike () was revealed, suggesting that the longer stimulus interval in Fig. 11B was one that resulted in Ca²⁺ spike augmentation, similar to the effect shown in Fig. 6C. In this cell, the low-threshold Na⁺ spike could be blocked by adequate somatic hyperpolarization (Fig. 11C). A small spike was then revealed (indicated by  in Fig. 11C) and blocked by further hyperpolarization.

View larger version (9K): [in this window] [in a new window]   FIG. 11. Dendritic Na+ spike influenced by somatic Na+ spike. All (oscilloscope) records in A-C from same cell. A and B: glutamate iontophoresis (3rd trace from top) on apical dendrite 407 µm from soma was adjusted to consistently evoke a low-threshold Na+ spike (top traces labeled 2) which was preceded by a brief current pulse (bottom traces) injected into the soma. Current pulse evoked a spike (top traces labeled 1) or was subthreshold for spike initiation (top traces labeled 2). Suprathreshold and subthreshold injected current pulse records in A and B are superimposed and spikes are truncated. A and B, 2nd traces from top: dV/dt of membrane potential used to mark spike initiation and show all spikes had same maximum rate of rise. A: iontophoretically evoked, low-threshold Na+ spike was occluded when somatic current injection evoked an action potential (trace 1), but was present if the somatic depolarization was just subthreshold (trace 2). B: when the interval between somatic and dendritic spikes was lengthened, a somatic action potential (trace 1) was followed by a small, broad spike, whereas the low-threshold Na+ spike followed a subthreshold somatic response (trace 2). C: when cell was hyperpolarized by injected current pulses (bottom trace), the low-threshold Na+ spike was replaced by a small, slow spike () that was abolished by further hyperpolarization. Calibrations: horizontal bar in A, 100 ms for A and B, 200 ms for C; vertical bar in A, 20 mV for A-C, 1,000 V/s for dV/dt records, 5 nA for injected current, 200 nA for iontophoretic current for all traces.

Low-threshold Na⁺ current spikes also were observed when the soma was voltage clamped during dendritic depolarization. In 16 cells current spikes were observed during iontophoresis when the soma was voltage clamped below the firing level of spikes evoked by somatic current injection, and the effects of TTX on these current spikes were tested. In seven of these cells, only TTX was tested and the low-threshold current spikes were abolished by addition of 1 µM TTX (data not shown). In nine other cells the effect of both TTX and Ca²⁺ channel blockade were tested, and each affected the current spikes as illustrated for one such cell in Fig. 12.

View larger version (25K): [in this window] [in a new window]   FIG. 12. Burst firing and dendritic Na+ spikes evoked by dendritic glutamate iontophoresis. All records from same cell. A: oscilloscope records of membrane potential response (top trace) evoked by current pulse (bottom trace) injected at soma. B: membrane potential response (top trace) evoked by iontophoresis (bottom trace) on dendrite 370 µm from soma. Spikes in A and B are truncated. Horizontal lines superimposed on membrane potential responses in A and B: records of membrane potential when iontophoresis was applied during DC voltage clamp of soma at indicated potentials (60 and 70 mV). C1-C4: digitized membrane currents (bottom traces) measured in voltage clamp during iontophoresis (top traces) at both holding potentials shown in A and B (C1) or at the depolarized holding potential (C2-C4) in control solutions (C1 and C3), when Mn2+ was substituted for Ca2+ in the perfusate (C2), and when 1 µM TTX was added (C4).

Figure 12 also illustrates the quite different responses evoked by somatic current injection and by dendritic depolarization that we observed in most cells that displayed a low-threshold, TTX-sensitive spike. Figure 12A shows arecord of the tonic repetitive firing evoked in this cell by constant current injected at the soma. No burst firing was evoked by injection of constant current at the soma up to 2 nA. In contrast, long-lasting dendritic depolarization evoked rhythmic bursts of action potentials (Fig. 12B). This contrasting response to somatic versus dendritic depolarization was observed in 12 of the 16 cells in which a low-threshold spike was evoked by dendritic depolarization and found to be TTX sensitive. Of these 12 cells, 10 exhibited no burst firing whatsoever during somatic current pulses up to 1-2 nA in amplitude, and the other 2 exhibited a single initial burst no longer than five action potentials. In 9 of these 12 cells only one iontophoretic strength was tested and it evoked one to three initial bursts of two to three action potentials followed by tonic firing that was similar to that evoked by somatic current injection (data not shown). In three cells (including that of Fig. 12) the effects of several iontophoretic strengths were examined. In these three cells rhythmic bursts, similar to those shown in Fig. 12B, occurred only at the minimal iontophoretic stimulus that evoked repetitive firing. Larger iontophoretic stimuli evoked a response similar to that seen in the other nine cells, namely, one to three initial bursts of two to three action potentials that were followed by tonic repetitive firing (data not shown).

The arrangement of Fig. 12 is similar to that of Fig. 3. The horizontal lines (labeled 60 mV and 70 mV) superimposed on the membrane potential responses in Fig. 12, A and B, are the actual membrane potential recorded during DC voltage clamp of the soma during iontophoresis. The records of Fig. 12, C1-C4, show the membrane current evoked by the iontophoresis at these holding potentials. Figure 12C1 shows that no current spike was evoked during iontophoresis in physiological saline when somatic membrane potential was held at resting potential (70 mV), but a large, early current spike was evoked when the iontophoresis was repeated with the soma held at 60 mV. That is, the current spike was evoked all-or-none by indirect alteration of dendritic membrane potential (similar to the procedure in Fig. 3, B and C). For clarity, only the glutamate-evoked currents at the depolarized holding potential (60 mV) are shown in Fig. 12 C2-C4, but this was the most negative holding potential at which a current spike appeared during iontophoresis. Note that this holding potential is well negative to the firing level of current-evoked spikes (Fig. 12A).

Substitution of Mn²⁺ for Ca²⁺ in the perfusate greatly reduced the amplitude of the current spike (cf. Fig. 12, C1-C3), indicating that Ca²⁺ channels were involved in generating the spike in physiological saline. The blockade of Ca²⁺ channels alone seems unlikely to cause such a large reduction of spike amplitude, however. Somatic membrane potential was not held perfectly constant during the generation of large current spikes in physiological saline, however. There was a transient depolarization (marked by * in Fig. 12A) associated with the current spike. It is possible that this brief somatic depolarization was sufficient to trigger a spike in the initial segment, and firing of the initial segment was responsible for generating some of the inward current during the large current spike observed in physiological saline. Somatic membrane potential was successfully held constant below somatic firing level both at the onset of the current spike and during the larger steady current after the current spike, however. Because a transient loss of voltage control requires the existence of a transient perturbing current, it is likely that the perturbing spike of current would have arisen in the dendrite even if some of the inward current during the spike were generated at the initial segment.

Evidence for the generation of a dendritic Na⁺ spike is based on the subsequent blockade of the Mn²⁺-resistant spike of Fig. 12C2 by TTX (Fig. 12C4). The membrane potential recorded during voltage clamp when the iontophoresis was repeated after substitution of Mn²⁺ for Ca²⁺ is superimposed on the burst response of Fig. 12B. (The burst response itself was recorded in physiological saline.) Somatic membrane potential was controlled when the Mn²⁺-resistant current spike of Fig. 12C2 was evoked. The reduction of the current spike by the Mn²⁺ substitution indicates Ca²⁺ involvement. The fact that a spike was still present in Mn²⁺ but eliminated by TTX indicates Na⁺ involvement. Because the TTX-sensitive spike was evoked by dendritic depolarization at a time when somatic membrane potential was successfully held constant below somatic firing level, the TTX-sensitive spike arose in the dendrite.

	DISCUSSION

Abstract Introduction Methods Results Discussion References

Our principal findings are that 1) depolarization of sites on the apical dendrite can initiate "local" dendritic Ca²⁺ spikes that do not actively propagate along the dendrite, 2) dendritic K⁺ channels normally prevent the active propagation of these Ca²⁺ spikes, and 3) dendritic depolarization can initiate dendritic Na⁺ spikes. Our observations also suggest that the dendritic Na⁺ spikes often are triggered by a local dendritic Ca²⁺ spike. When the iontophoretically evoked response did not reach somatic firing level, or when somatic membrane potential was voltage clamped below somatic firing level during dendritic depolarization, the dendritic spike activity inactivated during sustained dendritic depolarization. Another finding of interest was that many of the cells that exhibited dendritic spikes also responded with repetitive burst firing to long-lasting dendritic depolarization even though they exhibited no burst firing to somatic depolarization.

Several of our observations are consistent with the idea that the small, slow spikes triggered by dendritic depolarization were initiated in the dendrite. These spikes were triggered by dendritic but not somatic depolarization. They arose at somatic membrane potentials that were far below somatic firing level. They not only resisted hyperpolarization of the soma below resting potential, but their apparent threshold was then also below resting potential (Figs. 2C and 10C). Finally, these spikes were evoked by dendritic depolarization when spike initiation at the soma or axon was prevented by somatic voltage clamp. The low apparent threshold of these spikes, their small amplitude, and the decrease of their amplitude with increasing iontophoretic distance from the soma also argue for a dendritic origin. Their identification as Ca²⁺ spikes is based on their abolition by the partial or full substitution of other divalent cations for Ca²⁺ but not by the addition of TTX or NMDA channel blockers to the perfusate. Our evidence that these spikes arise in and are confined to a spatially restricted region of the dendrite is based on their small size, their low apparent threshold, the reduction of both their amplitude and their apparent threshold as the iontophoretic site was moved farther from the soma, and the appearance of large, low-threshold iontophoretically evoked Ca²⁺ spikes after K⁺ channels were blocked by TEA.

A theoretical model of dendritic K⁺ channels (Wilson 1995) has indicated that these K⁺ channels would isolate regenerative activity arising in a portion of the dendrite from events on the rest of the neuron. Our results from the TEA experiments are consistent with this idea, because the Ca²⁺ spikes were able to propagate actively to the soma when K⁺ channels were reduced by TEA but were confined to a local region of the dendrite when K⁺ channels were intact. We might expect the activation of dendritic K⁺ channels to cause the axial current transmitted from the depolarized site to be even more attenuated than in a passive dendrite (Wilson 1995). In contrast to this prediction, the transmission of tonic axial current to the soma was found to be better than expected for a passive dendrite (Schwindt and Crill 1995, 1996). The dendritic K⁺ channels were ineffective in preventing the observed amplification of tonic transmitted current, possibly because the tonic dendritic depolarization was not large enough to activate the dendritic K⁺ channels. Our present results suggest that a more important role of dendritic K⁺ channels is to prevent the active propagation of dendritic Ca²⁺ spikes along the dendrite.

It is simplest to suppose that the local Ca²⁺ spikes we observed were initiated at the site of iontophoresis, where the dendrite was most depolarized. Analysis of Ca²⁺ spike potentials (Reuveni et al. 1993) and Ca²⁺ imaging studies (Yuste et al. 1994) has suggested that Ca²⁺ spike generation (in the presence of TEA) arises preferentially from certain dendritic regions in neocortical pyramidal neurons. Some of these more excitable regions appear to be quite broad, however, and the excitability of these regions may be exaggerated by the blockade of K⁺ channels that normally limit excitability. Our most distal iontophoretic site was located only at the proximal edge of the dendritic region where increased Ca²⁺ accumulation was observed in imaging studies (Yuste et al. 1994), however. Our results therefore shed no light on Ca²⁺ spike initiation in the distal half of the apical dendrite. It is possible that the Ca²⁺ spikes we observed can be initiated only at discrete dendritic sites, but these sites must be numerous and closely spaced. Because the local Ca²⁺ spikes were observed over the whole range of iontophoretic distances tested in our study (278-555 µm), much or all of the apical dendrite in this region must be capable of generating the local Ca²⁺ spikes. This idea is consistent with the implication of imaging studies that voltage-gated Ca²⁺ channels exist on the entire dendritic tree. Our observations may also have some relevance to the surprisingly large increases of intracellular Ca²⁺ concentration that were observed to accompany subthreshold, non-NMDA-mediated EPSPs (Markram and Sakmann 1994). The large increase of intracellular Ca²⁺ concentration could arise if these EPSPs evoke Ca²⁺ spikes that are restricted to the region near the synaptic input. Given the small amplitude of the local Ca²⁺ spikes that we observed, it may be difficult to distinguish a pure EPSP from an EPSP with a small superimposed Ca²⁺ spike.

Dendritic depolarization also evoked low-threshold Na⁺ spikes in many recorded cells. Our evidence that these Na⁺ spikes were initiated in the dendrite is similar to our evidence for the Ca²⁺ spikes and is based on several observations. The apparent firing level of these spikes was far below that of spikes evoked by somatic depolarization, which are presumably triggered at the initial segment. They could survive somatic hyperpolarization, in which case their apparent threshold was negative to resting potential (Fig. 10A). These low-threshold Na⁺ spikes were evoked only by dendritic depolarization, never by somatic depolarization. When the soma was voltage clamped at potentials that prevented initiation of a spike in the soma or axon, TTX-sensitive current spikes were evoked by dendritic depolarization (Fig. 12). Thus our observations on the initiation of these Na⁺ spikes parallel our observations on the Ca²⁺ spikes and point to a dendritic site of initiation. Our evidence that the spikes are Na⁺ spikes is based on their abolition by TTX. These low-threshold Na⁺ spikes differed from the Ca²⁺ spikes in size and duration. They were similar in duration and amplitude to the spikes evoked by somatic current injection (Fig. 8, B1 and B2). Thus, unlike the Ca²⁺ spikes, they propagated actively to the soma after originating in the dendrite.

Our observation of local dendritic Ca²⁺ spikes may reconcile conflicting ideas about dendritic Na⁺ spike initiation. Several other investigators have obtained evidence for the dendritic initiation of Na⁺ spikes in both hippocampal pyramidal neurons (Colling and Wheal 1994; Poolos and Kocis 1990; Turner et al. 1991; Wong and Stewart 1992) and layer 5 pyramidal neurons (Kim and Connors 1993; Regehr et al. 1993). Other investigators, in contrast, have provided strong evidence that the Na⁺ spike is normally initiated downstream from the soma and subsequently invades the dendrites (Stuart and Sakmann 1994). A recent theoretical study (Mainen et al. 1995) suggested that the dendrites cannot initiate Na⁺ spikes because the local dendritic membrane potential rises too slowly as a consequence of low dendritic Na⁺ channel density. According to this study the dendritic Na⁺ current inactivates during the slow rise of dendritic membrane potential, and regenerative depolarization cannot be initiated. The back-propagating spike, in contrast, depolarizes dendritic membrane potential rapidly enough to avoid dendritic Na⁺ inactivation. This model did not include a noninactivating dendritic Na⁺ current, however. If the dendrites generate such a noninactivating inward current (Schwindt and Crill 1995, 1996), inactivation of the transient Na⁺ current would be less relevant. In addition, when we eliminated the low-threshold Na⁺ spike by electrophysiological (Fig. 11, B and C) or pharmacological means (Fig. 10C), we usually found an underlying Ca²⁺ spike. These observations lead us to propose that the dendritic Na⁺ spike is normally triggered by the dendritic Ca²⁺ spike. When evoked by adequate excitatory input, the local Ca²⁺ spike may provide the initial, rapid depolarization of dendritic membrane potential that allows a regenerative Na⁺ spike to develop before inactivation ensues. It is possible that the most important electrophysiological function of the local Ca²⁺ spike is to trigger a dendritic Na⁺ spike. This process would be facilitated if dendritic K⁺ currents were reduced. When dendritic K⁺ currents were reduced by TEA, the dendritic Ca²⁺ spike itself actively propagated to the soma. It is possible that the dendritic K⁺ currents can be reduced (modulated) by certain neurotransmitters acting through second messengers. This may enhance local dendritic Ca²⁺ spikes and thereby facilitate the initiation of a dendritic Na⁺ spike.

It is possible that a Na⁺ spike is initiated directly in some dendrites or under some conditions without an underlying Ca²⁺ spike. This idea seems to be supported by our results from those cells studied in voltage clamp, in which the low-threshold current spikes were entirely abolished by TTX. In other cells, Ca²⁺ channel blockade reduced the low-threshold current spike (e.g., Fig. 12C2), implying an additional Ca²⁺-dependent component of the spike. But if dendritic Na⁺ spikes are normally triggered by local dendritic Ca²⁺ spikes as we propose, why was a residual Ca²⁺ spike not seen in the presence of TTX in these cells (e.g., Fig. 12C4)? A scenario that reconciles these observations with the idea that a Ca²⁺ spike normally triggers the dendritic Na⁺ spike is as follows. We have shown previously that the dendrites also possess a noninactivating Na⁺ current, I_NaP (Schwindt and Crill 1995, 1996), which is activated by a smaller depolarization than required for a Ca²⁺ spike. Thus dendritic depolarization would first activate dendritic I_NaP, which may then provide the additional depolarization required to trigger a Ca²⁺ spike, which would cause the additional, rapid depolarization required to trigger a regenerative Na⁺ spike. TTX application would block both I_NaP and transient Na⁺ current. According to this idea, a residual Ca²⁺ spike was not observed in those cells in which TTX totally abolished the low-threshold spike because TTX also blocked I_NaP. In the absence of I_NaP, dendritic depolarization would remain below the threshold for a regenerative Ca²⁺ spike. It is possible, therefore, that dendritic Na⁺ spikes always are evoked by local Ca²⁺ spikes even though we observed that TTX completely eliminated low-threshold spike activity in some cells.

The primary reason that we used glutamate iontophoresis in these experiments was that it provided a convenient way to depolarize a site on the dendrite. Dendritic depolarization by a method other than electrically evoked EPSPs was required to study the ionic basis of the spike responses. The identification of ionic mechanisms required the application of agents that would block or alter synaptic transmission (TEA, TTX, divalent cations substituted for Ca²⁺, NMDA-receptor blockers). The use of glutamate iontophoresis had other advantages, however. It allowed us to reliably and conveniently grade the amplitude and duration of dendritic excitation, which would be difficult to do with the use of electrically evoked synaptic stimulation. It provided a smooth, graded postsynaptic response on which it was easy to recognize and test for a small, all-or-none spike. Asynchronous excitatory synaptic input is particularly difficult to mimic with the use of electrical stimulation, and the dropout of a transient response at a lower stimulus intensity could result simply from failure to stimulate a subset of afferent fibers rather than the failure of a small, postsynaptic action potential.

Glutamate is widely accepted as the primary neurotransmitter responsible for fast dendritic EPSPs in these cells. It is simplest to suppose that the iontophoresis of glutamate in our experiments depolarized the dendrite by opening glutamate-activated channels, just as excitatory synaptic input does, and thereby mimicked tonic, asynchronous, glutamatergic synaptic input to the dendritic site. It is possible, however, that there are significant differences between the effect of glutamate iontophoresis and glutamatergic synaptic input. These differences could include excessive glutamate stimulation, which might abnormally depolarize the dendrite and allow excessive Ca²⁺ entry, stimulation of extrasynaptic receptors, stimulation of metabotropic as well as ionotropic receptors, and the local release of other neuroactive substances due to stimulation of nearby presynaptic cells, fibers, or terminals. These reservations apply, of course, to all studies that have used glutamate iontophoresis to excite neurons.

Concerning the possibility of excessive glutamate stimulation, the best indicator of the intensity of the input is the intensity of the output. Most iontophoresis resulted in somatic depolarizations below firing level, and tonic repetitive firing was evoked only at low rates (<30 Hz). (This excludes the high-rate spikes during bursts, but bursting subsided to tonic firing as iontophoretic strength was increased.) Many identified pyramidal tract neurons in motor cortex of awake, behaving monkeys have been observed to fire at rates 50 Hz for several seconds during the performance of motor tasks (e.g., Cheney and Fetz 1980). Therefore our glutamatergic stimulation appears to have been mild compared with what pyramidal neurons experience during natural activation, assuming that most input during natural activation is glutamatergic.

The danger of stimulating extrasynaptic receptors, metabotropic receptors, or presynaptic elements releasing other neuroactive substances is that intracellular biochemical pathways might be activated in the recorded cell that would alter its membrane properties. Then, dendritic spikes may have arisen not simply because the dendrites were adequately depolarized but also because certain membrane properties were altered that are not altered by stimulation of subsynaptic glutamate receptors. There are several lines of evidence suggesting that these potentially confounding factors have only a minor influence, if any, on our results. For example, there is no reason to think that these factors would be less for iontophoresis on the soma, but the unusual spikes we observed were evoked only by iontophoresis on the dendrites. We believe the iontophoretically evoked postsynaptic response is mediated almost exclusively by -amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid (AMPA)- and NMDA-sensitive glutamate receptors, because we found the postsynaptic response to be abolished (reversibly) by the combined bath application of 6-cyano-7-nitroquinoxaline-2,3-dione (CNQX) and APV (50 µM each, n = 2) or CNQX alone (50 µM; n = 1) (unpublished observations). Bath application of specific metabotropic agonists to these cells evoked only minor depolarization and responses quite unlike those reported here (Greene et al. 1994), and the described metabotropic responses were not seen in this study. Concerning stimulation of presynaptic cells or fibers, we have found consistently that TTX application (which would block conduction in presynaptic cells or fibers) does not alter the amplitude of the response at potentials at which voltage-gated dendritic channels are expected to be closed (Schwindt and Crill 1995, 1996), and we observed dendritic Ca²⁺ spikes in the presence of TTX in the present study (Figs. 4C, 5, and 10C). Concerning release of substances from presynaptic terminals, we observed dendritic Na⁺ spikes after the blockade of Ca²⁺ channels in the postsynaptic membrane and (presumably) on presynaptic terminals (Fig. 12, C2-C4).

It thus seems likely that the glutamate iontophoresis depolarized the dendrite predominantly in the expected way, by opening postsynaptic glutamate-gated channels, and the dendritic depolarization evoked the dendritic spikes. Insofar as this depolarization mimics the effect of glutamatergic synaptic input, it provides a more natural method of examining input-output properties than somatic current injection, the stimulus that traditionally has been used to investigate the firing properties of these neurons. On this assumption, we interpret our findings in terms of how the neuron may respond to tonic, excitatory, dendritic synaptic input.

The local Ca²⁺ spikes clearly provide a boost to local dendritic depolarization, because Ca²⁺ spike amplitude viewed from the soma was greater than the underlying graded depolarization. The boosting provided by the local Ca²⁺ spikes is limited, however. It is available only during the onset of a depolarization from resting potential, and Ca²⁺ spike amplitude decreases with distance from the soma, similar to the decrement of EPSP amplitude with distance expected in a passive dendrite. Thus the Ca²⁺ spike would enhance the effect of excitatory input but would not radically alter the relative weighting of more distal excitatory input. The neuron would still largely be integrative, because excitatory input from a number of afferent fibers impinging at a dendritic site might be required to trigger a Ca²⁺ spike at that site, and several Ca²⁺ spikes from different locations (especially in the distal dendrites) might be needed to trigger a spike at the initial segment. Furthermore, the Ca²⁺ spikes might occur only during the onset of tonic excitatory input.

If excitatory input triggers a dendritic Na⁺ spike, the neuron would be integrative only in the sense that input from several afferents may be needed to reach the local threshold (possibly, the threshold for evoking a local Ca²⁺ spike), but whenever local membrane potential exceeded local threshold a Na⁺ spike would propagate through the cell and down the axon. Because of the high input resistance of small-diameter cylinders, this response mode could cause the distal dendrites (where an excitatory synaptic current would most readily cause a suprathreshold depolarization) to become the most excitable part of the cell and to dominate the cell's output. Again, these dendritic Na⁺ spikes might occur only at the onset of tonic excitatory input.

The interaction of somatically initiated Na⁺ spikes with dendritic Ca²⁺ and Na⁺ spikes (Figs. 6 and 11) further complicates our ideas of neural integration. One implication of this finding is that the dendritic spike mechanism is unavailable if a somatic Na⁺ spike has occurred earlier within a critical interval. If evoked after a longer interval, the somatic Na⁺ spike may be followed by a larger-than-normal Ca²⁺ spike (Fig. 6C), although not necessarily by a dendritic Na⁺ spike (Fig. 11B). These observations are consistent with the back-propagation of the somatic Na⁺ spike to the iontophoretic site, because the dendritic membrane is expected to be less excitable after invasion of a back-propagated spike. One cause of this reduced excitability may simply be membrane potential hyperpolarization if the dendritic membrane generates a postspike afterhyperpolarization (Andreasen and Lambert 1995). The enhancement of the dendritic Ca²⁺ spike in Fig. 6C is consistent with the removal of inactivation of Ca²⁺ channels at the dendritic site, perhaps by an afterhyperpolarization associated with a back-propagated Na⁺ spike. The only firm conclusion we can draw from these observations, however, is that complicated interactions are possible between somatically initiated and dendritically initiated spikes. When the dendritic input is strong enough to cause repetitive firing, these interactions may lead to responses that cannot be foreseen from the behavior of subthreshold responses.

One such unforeseen response was the rhythmic burst firing that we observed during low-strength iontophoresis in some cells. Each of these cells exhibited low-threshold, TTX-sensitive current spikes during voltage clamp, but the current spikes were seen only at the onset of the iontophoresis during voltage-clamp recording. What then could be responsible for repetitive burst firing observed throughout the iontophoresis in current-clamp recording? A speculative explanation is based on the enhancement of the Ca²⁺ spike that we observe after a single somatic Na⁺ spike: it is possible that the large afterhyperpolarization that followed a burst of Na⁺ spikes resulted in a large, "rebound" dendritic Ca²⁺ spike that triggered another burst of Na⁺ spikes, etc. The rhythmic bursts were not seen at higher iontophoretic strength (i.e., dendritic depolarization), possibly because the dendritic spike mechanisms inactivated after the initial bursts.

A final implication of our study is that the input-output relation of neocortical pyramidal neurons to a constant excitatory input is not entirely revealed by intrasomatic injection of a constant current. This has been demonstrated by simultaneous somatic and dendritic recording and current injection in hippocampal pyramidal neurons (Wong and Stewart 1992). Our results suggest that the same conclusion can be drawn when the dendrite is depolarized by activation of glutamate-gated channels. We have shown elsewhere that somatic current injection usually gives an accurate picture of the steady-state response of these neurons to tonic or slowly changing dendritic glutamatergic input (Schwindt and Crill 1996), but the present results show that somatically injected current does not adequately reveal the cell's repertoire of responses to steady glutamatergic input in those cells capable of generating dendritic Na⁺ spikes and the concomitant rhythmic burst firing. In no cell in this study were the subthreshold oscillations due to dendritic Ca²⁺ spikes or the low-threshold Na⁺ spikes mimicked by somatic current injection. Even in those cells that exhibited an initial burst at the onset of an intrasomatic current pulse, subthreshold Ca²⁺ spike oscillations and low-threshold Na⁺ spikes were absent. On the other hand, cells that exhibited no burst firing whatsoever to somatic depolarization did show initial membrane potential oscillations or bursts of action potentials during dendritic depolarization. On the basis of their response to somatically injected current, rodent neocortical neurons have been separated into two general classes, intrinsic bursters or regular-spiking nonbursters (Connors and Gutnick 1990). Our present results suggest that this division may be a function of the location of the depolarizing stimulus and therefore somewhat artificial: "nonbursters" defined by somatic injected current often exhibited transient or rhythmic burst firing during dendritic depolarization.

In summary, our experiments show that the apical dendrite of neocortical pyramidal neurons can transform local glutamatergic input in three ways: by amplifying the tonic axial current that is transmitted to the soma (Schwindt and Crill 1995, 1996); by generating local Ca²⁺ spikes that further, transiently, amplify the input signal; and by the initiation of dendritic Na⁺ spikes that propagate to the soma.

	ACKNOWLEDGEMENTS

  We thank G. Hinz and P. Newman for excellent technical assistance.

  This work was supported by National Institute of Neurological Disorders and Stroke Grant NS-16792 and the Keck Foundation.

	FOOTNOTES

  Address for reprint requests: P. C. Schwindt, Dept. of Physiology and Biophysics, University of Washington School of Medicine, Box 357290, Seattle, WA 98195-7290.

  Received 5 June 1996; accepted in final form 14 January 1997.

	REFERENCES

Abstract Introduction Methods Results Discussion References

• AMITAI, Y., FRIEDMAN, A., CONNORS, B. W., GUTNICK, M. J. Regenerative activity in apical dendrites of pyramidal cells in neocortex. Cereb. Cortex 3: 26-38, 1993.[Abstract/Free Full Text]
• ANDREASEN, M., LAMBERT, J.D.C. The excitability of CA1 pyramidal cell dendrites is modulated by a local Ca²⁺-dependent K⁺-conductance. Brain Res. 698: 193-203, 1995.[Medline]
• BENARDO, L. S., MASUKAWA, L. M., PRINCE, D. A. Electrophysiology of isolated hippocampal pyramidal dendrites. J. Neurosci. 2: 1614-1622, 1982.[Abstract]
• CHENEY, P. D., FETZ, E. E. Functional classes of primate corticomotoneuronal cells and their relation to active force. J. Neurophysiol. 44: 773-791, 1980.[Free Full Text]
• COLBERT, C. M., JOHNSTON, D. Axonal action potential initiation and Na⁺ channel densities in the soma and axon initial segment of subicular pyramidal neurons. J. Neurosci. 16: 6676-6689, 1996.[Abstract/Free Full Text]
• COLLING, S. B., WHEAL, H. V. Fast sodium action potentials are generated in the distal apical dendrites of rat hippocampal CA1 pyramidal cells. Neurosci. Lett. 172: 73-76, 1994.[Medline]
• CONNORS, B. W., GUTNICK, M. J. Intrinsic firing patterns of diverse neocortical neurons. Trends Neurosci. 13: 99-104, 1990.[Medline]
• ECCLES, J. C. In: The Physiology of Nerve Cells., . Baltimore, MD: Johns Hopkins Press, 1957
• FLATMAN, J. A., SCHWINDT, P. C., CRILL, W. E. The induction and modification of voltage-sensitive responses in cat neocortical neurons by N-methyl-D-aspartate. Brain Res. 363: 62-77, 1986.[Medline]
• GREENE, C., SCHWINDT, P., CRILL, W. E. Properties and ionic mechanisms of a metabotropic glutamate receptor-mediated afterdepolarization in neocortical neurons. J. Neurophysiol. 72: 693-704, 1994.[Abstract/Free Full Text]
• HU, G.-Y., HVALBY, Ø. Glutamate-induced action potentials are preceded by regenerative prepotentials in rat hippocampal pyramidal cells in vitro. Exp. Brain Res. 88: 485-494, 1992.[Medline]
• JAFFE, D. B., JOHNSTON, D., LASSER-ROSS, N., LISMAN, J. E., MIYAKAWA, H., ROSS, W. N. The spread of Na⁺ spikes determines the pattern of dendritic Ca²⁺ entry into hippocampal neurons. Nature Lond. 357: 244-246, 1992.[Medline]
• KIM, H. G., CONNORS, B. W. Apical dendrites of the neocortex: correlation between sodium- and calcium-dependent spiking and pyramidal cell morphology. J. Neurosci. 13: 5301-5311, 1993.[Abstract]
• LLINAS, R., SUGIMORI, M. Electrophysiological properties of in vitro Purkinje cell dendrites in mammalian cerebellar slices. J. Physiol. Lond. 305: 197-213, 1980.[Abstract/Free Full Text]
• MAGEE, J. C., CHRISTOFI, G., MIYAKAWA, H., CHRISTIE, B., LASSER-ROSS, N., JOHNSTON, D. Subthreshold synaptic activation of voltage-gated Ca²⁺ channels mediates a localized Ca²⁺ influx into the dendrites of hippocampal pyramidal neurons. J. Neurophysiol. 74: 1335-1342, 1995.[Abstract/Free Full Text]
• MAGEE, J. C., JOHNSTON, D. Synaptic activation of voltage-gated channels in the dendrites of hippocampal pyramidal neurons. Science Wash. DC 268: 301-304, 1995a.[Abstract/Free Full Text]
• MAGEE, J. C. AND JOHNSTON, D. Characterization of single voltage-gated Na⁺ and Ca²⁺ channels in apical dendrites of rat pyramidal neurons. J. Physiol. Lond. 487: 1: 67-90, 1995b.
• MAINEN, Z. F., JORGES, J., HUGUENARD, J. R., SEJNOWSKI, T. J. A model of spike initiation in neocortical pyramidal neurons. Neuron 15: 1427-1439, 1995.[Medline]
• MARKRAM, H., HELM, P. J., SAKMANN, B. Dendritic calcium transients evoked by single back-propagating action potentials in rat neocortical pyramidal neurons. J. Physiol. Lond. 485: 1-20, 1995.[Abstract/Free Full Text]
• MARKRAM, H., SAKMANN, B. Calcium transients in dendrites of neocortical neurons evoked by single subthreshold excitatory postsynaptic potentials via low-voltage-activated calcium channels. Proc. Natl. Acad. Sci. USA 91: 5207-5211, 1994.[Abstract/Free Full Text]
• MIYAKAWA, H., LEV-RAM, V., LASSER-ROSS, N., ROSS, W. N. Calcium transients evoked by climbing fiber and parallel fiber synaptic inputs in guinea pig cerebellar Purkinje neurons. J. Neurophysiol. 68: 1178-1189, 1992.[Abstract/Free Full Text]
• POCKBERGER, H. Electrophysiological and morphological properties of rat motor cortex neurons in vivo. Brain Res. 539: 181-190, 1991.[Medline]
• POOLOS, N. P., KOCIS, J. D. Dendritic action potentials activated by NMDA receptor-mediated EPSPs in CA1 hippocampal pyramidal cells. Brain Res. 524: 342-346, 1990.[Medline]
• , sect. 1, vol. I, part 1, chapt. 3 p. 39-97. RALL, W. Core conductor theory and cable properties of neurons. In: Handbook of Physiology. The Nervous System. Cellular Biology of Neurons., . Bethesda, MD: Am. Physiol. Soc., 1977
• RALL, W., SEGEV, I. Space-clamp problems when voltage clamping branched neurons with intracellular microelectrodes. In: Voltage and Patch Clamping with Microelectrodes, edited by T. G. Smith Jr., H. Lecar, S. J. Redman, and P. W. Gage . Bethesda, MD: Am. Physiol. Soc., 1985, p. 191-216
• REGEHR, W. G., CONNOR, J. A., TANK, D. W. Optical imaging of calcium accumulation in hippocampal pyramidal cells during synaptic activation. Nature Lond. 341: 533-536, 1989.[Medline]
• REGEHR, W. G., KEHOW, J., ASCHER, P., ARMSTRONG, C. Synaptically triggered action potentials in dendrites. Neuron 11: 145-151, 1993.[Medline]
• REUVENI, I., FRIEDMAN, A., AMITAI, Y., GUTNICK, M. J. Stepwise repolarization from Ca²⁺ plateaus in neocortical pyramidal cells: evidence for nonhomogeneous distribution of HVA Ca²⁺ channels in dendrites. J. Neurosci. 13: 4609-4621, 1993.[Abstract]
• ROSS, W. N., LASSER-ROSS, N., WERMAN, R. Spatial and temporal analysis of calcium-dependent electrical activity in guinea pig Purkinje cell dendrites. Proc. R. Soc. Lond. B Biol. Sci. 240: 173-185, 1990.[Medline]
• SCHILLER, J., HELMCHEN, F., AND SAKMANN, B. Spatial profile of dendritic calcium transients evoked by action potentials in rat neocortical pyramidal neurones. J. Physiol. Lond. 487: 3: 583-600, 1995.
• SCHWINDT, P. C., CRILL, W. E. Amplification of synaptic current by persistent sodium conductance in apical dendrite of neocortical neurons. J. Neurophysiol. 74: 2220-2224, 1995.[Abstract/Free Full Text]
• SCHWINDT, P. C., CRILL, W. E. Equivalence of amplified current flowing from dendrite to soma measured by alteration of repetitive firing and by voltage clamp in layer 5 pyramidal neurons. J Neurophysiol. 76: 3731-3739, 1996.[Abstract/Free Full Text]
• SPENCER, W. A., KANDEL, E. R. Electrophysiology of hippocampal neurons. IV. Fast prepotentials. J. Neurophysiol. 24: 272-285, 1961.[Free Full Text]
• STAFSTROM, C. E., SCHWINDT, P. C., CHUBB, M. C., CRILL, W. E. Properties of persistent sodium conductance and calcium conductance of layer V neurons from cat sensorimotor cortex in vitro. J. Neurophysiol. 53: 152-170, 1985.
• STAFSTROM, C. E., SCHWINDT, P. C., FLATMAN, J. A., CRILL, W. E. Properties of the subthreshold response and action potential recorded in layer V neurons from cat sensorimotor cortex in vitro. J. Neurophysiol. 52: 244-263, 1984.[Abstract/Free Full Text]
• STUART, G. J., SAKMANN, B. Active propagation of somatic action potentials into neocortical pyramidal cell dendrites. Nature Lond. 367: 69-72, 1994.[Medline]
• TURNER, R. W., MEYERS, D.E.R., RICHARDSON, T. L., BARKER, J. L. The site for initiation of action potential discharge over the somatodendritic axis of rat hippocampal CA1 pyramidal neurons. J. Neurosci. 11: 2270-2280, 1991.[Abstract]
• WILSON, C. J. Dynamic modification of dendritic cable properties and synaptic transmission by voltage-gated potassium channels. J. Comput. Neurosci. 2: 91-115, 1995.[Medline]
• WONG, R.K.S., PRINCE, D. A. Participation of calcium spikes during intrinsic burst firing in hippocampal neurons. Brain Res. 159: 385-390, 1978.[Medline]
• WONG, R.K.S., STEWART, M. Different firing patterns generated in dendrites and somata of CA1 pyramidal neurones in guinea-pig hippocampus. J. Physiol. Lond. 457: 675-687, 1992.[Abstract/Free Full Text]
• YUSTE, R., GUTNICK, M. J., SAAR, D., DELANEY, K. R., TANK, D. W. Ca²⁺ accumulations in dendrites of neocortical pyramidal neurons: an apical band and evidence for two functional compartments. Neuron 13: 23-43, 1994.[Medline]
• ZILLES, K., WREE, A. Cortex: areal and laminar structure. In: The Rat Nervous System, edited by and G. Paxinos . Kensington, Australia: Academic, 1985, vol. 1, p. 375-415

0022-3077/97 $5.00 Copyright ©1997 The American Physiological Society