Distal Initiation and Active Propagation of Action Potentials in Interneuron Dendrites

See allHide authors and affiliations

Science  14 Jan 2000:
Vol. 287, Issue 5451, pp. 295-300
DOI: 10.1126/science.287.5451.295


Fast and reliable activation of inhibitory interneurons is critical for the stability of cortical neuronal networks. Active conductances in dendrites may facilitate interneuron activation, but direct experimental evidence was unavailable. Patch-clamp recordings from dendrites of hippocampal oriens-alveus interneurons revealed high densities of voltage-gated sodium and potassium ion channels. Simultaneous recordings from dendrites and somata suggested that action potential initiation occurs preferentially in the axon with long threshold stimuli, but can be shifted to somatodendritic sites when brief stimuli are applied. After initiation, action potentials propagate over the somatodendritic domain with constant amplitude, high velocity, and reliability, even during high-frequency trains.

γ-Aminobutyric acid (GABA)–containing interneurons control the activity of cortical neuronal networks (1). Interneurons mediate feedback and feedforward inhibition (2), set the threshold for initiation of axonal Na+ action potentials and dendritic Ca2+ spikes in principal neurons (3), and participate in the generation of oscillatory activity (4). In many circuits, interneurons operate as coincidence detectors or relays that are activated with very short delay by a small number of principal neurons (5). The mechanisms that underlie the speed and efficacy of interneuron activation have not been determined yet. One possible factor is the rapid time course and the large amplitude of the α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid receptor-mediated conductance at principal neuron-interneuron synapses (6). Theoretical considerations, however, indicate that the properties of excitatory input are insufficient to explain fast and reliable input-output transformation and suggest the presence of active conductances in interneuron dendrites (7).

To address whether interneuron dendrites are active, we recorded from the dendrites of horizontal oriens-alveus interneurons of the hippocampal CA1 region (Fig. 1A) (8–10). We first examined the morphology of these interneurons, using biocytin as an intracellular marker. All interneurons examined were immunopositive for somatostatin, suggesting that they represented a neurochemically homogeneous population (10 neurons) (Fig. 1B) (1, 8). In most cells, the axon originated from one of the dendrites, up to 110 μm from the soma, and often terminated in stratum lacunosum-moleculare (105 neurons total) (Fig. 1, C and D) (11), consistent with previous descriptions (8). Electron microscopy combined with GABA immunostaining indicated that dendrites were densely covered with excitatory and inhibitory synapses (Fig. 1E). Quantitative analysis revealed that the density of synapses per surface area increased from the soma to the distal dendrites (Fig. 1F). These results suggest that the dendrites of oriens-alveus interneurons have a dual role, operating as sites for synaptic input and axonal output.

Figure 1

Interneuron dendrites are the sites of synaptic input and axonal output. (A) Infrared differential interference contrast video image of a dendrite of an oriens-alveus interneuron. The patch pipette is placed at a distance of 75 μm from the center of the soma. (B) Somatostatin immunoreactivity of a biocytin-filled oriens-alveus interneuron shown by fluorescent double labeling. (C) Camera lucida reconstruction of a biocytin-filled oriens-alveus interneuron. Axon, red; soma and dendrites, black. Str. l.-m., stratum lacunosum-moleculare; str. rad., stratum radiatum; str. pyr., stratum pyramidale; str. ori., stratum oriens. (D) Histogram of the distance of the axon initial segment (AIS) from the center of the soma. Sub., subiculum. (E) Electron microphotograph of synapses on the axon-bearing dendrite of an oriens-alveus interneuron (65 μm from the soma). D, dendrite; s, spinelike structure; b1, GABA-immunopositive bouton; b2, GABA-immunonegative (putative excitatory) bouton. Arrowheads indicate synaptic clefts. (F) Density of all synapses (three cells) and GABAergic synapses (solid portions of bars, two cells) on five selected regions of oriens-alveus interneurons. Distal dendrites are ≥100 μm from the soma, and proximal dendrites are ≤65 μm from the soma. Error bars represent SEM.

To examine whether the dendrites of oriens-alveus interneurons contained voltage-gated conductances, we isolated outside-out patches from different regions of the somatodendritic domain (Fig. 2) (12). In all cases, the location of the axon initial segment was determined by subsequent staining. Voltage pulses to −10 mV evoked a fast Na+inward current that was blocked by 500 nM tetrodotoxin (TTX) (Fig. 2A), followed by a slower K+ outward current that was reduced by 1 or 10 mM tetraethylammonium (TEA) or 3 mM 4-aminopyridine in dendritic and somatic patches. Conductance densities for Na+ and K+ were high and uniform over the somatodendritic domain, with no significant differences between the soma, axon-bearing dendrite, and axon-lacking dendrite (P > 0.2; 53 dendritic and 27 somatic patches) (Fig. 2B).

Figure 2

Active conductances in dendrites of oriens-alveus interneurons. (A) Na+ and K+ currents in a dendritic outside-out patch. Current in control conditions, in the presence of 500 nM external TTX, and Na+ current obtained by digital subtraction. Prepulse is to −120 mV (50 ms), and test pulse is to −10 mV. (B) Na+ (solid symbols and continuous lines) and K+(open symbols and dashed lines) current densities (I Na and I K, respectively), calculated from maximal inward or outward current at −10 mV (15) and plotted against distance from the center of the soma (positive values indicate the axon-bearing dendrite). Four patches in which neither Na+ nor K+ current could be evoked (probably due to vesicle formation) were excluded. (C) Na+channel activation curves. Solid symbols, dendritic channels (11 patches); open symbols, somatic channels (6 patches).P Na, Na+permeability; P Na max, maximal Na+ permeability; V, voltage. Inset shows corresponding Na+ currents in dendritic outside-out patch. Data were fitted with Boltzmann functions raised to the third power (midpoint potentials are −45.6 and −37.8 mV, respectively). Prepulses are to −120 mV (50 ms). Error bars in (C) through (F) represent SEM. (D) Recovery of Na+ channels from inactivation. The 30-ms pulses are to −10 mV; the interpulse potential is −120 mV. Inset shows corresponding Na+ inward currents during first and second pulse for interpulse intervals <10 ms. Line represents the exponential function fitted to the data points (time constant of 5.1 ms, five dendritic patches). (E) K+ channel activation curves. Solid symbols, dendritic channels; open symbols, somatic channels (12 patches in both cases).G K, K+ conductance;G K max, maximal K+ conductance. Inset shows corresponding K+ outward currents in dendritic outside-out patch. Data were fitted with Boltzmann functions raised to the fourth power (midpoint potentials are −15.3 and −15.6 mV, respectively). (F) Effect of 10 mM TEA on patches from the axon-lacking dendrite, soma, and axon-bearing dendrite (5, 6, and 8 patches, respectively). The 100-ms pulses are to 70 mV.I TEA, peak K+ current in the presence of TEA; I K tot, total K+current in control conditions. The ratio of steady state to peak current was 0.75 ± 0.05 (dendrite) and 0.66 ± 0.07 (soma) in the control and 0.52 ± 0.07 (dendrite) and 0.48 ± 0.12 (soma) in 10 mM TEA.

We next compared the gating properties of dendritic and somatic channels. The midpoint potential of Na+ channel activation was more negative in dendritic patches (Fig. 2C) (13). The time constant of recovery of Na+ channels from inactivation was fast, independent of location [5.1 ms for dendritic patches (Fig. 2D) and 5.3 ms for three somatic patches]. The midpoint potential of K+ channel activation was almost identical in dendritic and somatic patches (Fig. 2E). Both TEA-sensitive sustained currents and TEA-resistant A currents were present in oriens-alveus interneurons (9). Unlike in principal neurons (14), however, the ratio of the two K+ current components was not significantly different between dendrite and soma (Fig. 2F). Thus, the functional properties of active conductances in dendrites and somata were relatively similar.

The peak Na+ conductance density in oriens-alveus interneuron dendrites at –10 mV was 113 ± 9 pS μm−2 (15), comparable to that in dendrites of mitral cells in the olfactory bulb (90 pS μm−2) (16) but about three times that in dendrites of cortical principal neurons (40 pS μm−2) (17). This suggests that the processes of action potential initiation and propagation in oriens-alveus interneurons may differ from those in principal cells (17, 18). We therefore made simultaneous recordings from somata and dendrites (up to 110 μm from the soma) (Fig. 3) (12). Sustained, low-intensity current injection evoked action potentials that were detected first at the dendritic site in 10 of 25 cells (Fig. 3A, top traces) and at the somatic site in 15 cells (Fig. 3A, bottom traces), independent of the site of injection. In all neurons, subsequent morphological analysis revealed that the site where the action potential was recorded earlier (the preferred site) was closer to the axon initial segment (Fig. 3A, schemes on the left). This suggests axonal initiation, as described previously for neurons in the substantia nigra that show comparable morphology (19). However, when brief high-intensity stimuli were applied to the nonpreferred site, the temporal sequence of the action potentials was reversed (12 of 19 cells) or the delay between the action potentials was reduced (5 of 19 cells) (Fig. 3, B and C). Thus, for sustained low-intensity stimulation, the default action potential initiation site appears to be the axon, whereas for brief high-intensity stimulation, the initiation site shifts to nonpreferred dendritic or somatic locations.

Figure 3

Multiple action potential initiation sites in oriens-alveus interneurons. (A) Simultaneous current-clamp (CC) recordings from a dendrite (red traces) and soma (black traces) in two different neurons (middle panel, top and bottom). First action potentials in a train are shown on an expanded time scale in the right panel. Schemes on the left illustrate the locations of axon initial segments as revealed by correlated morphological analysis. The depolarizing current (100 ms, 100 pA) was injected into the soma in both cases. (B) The initiation site, which was close to the dendritic recording site with a long low-intensity pulse (upper traces), was shifted toward the somatic recording site with a brief pulse of high intensity [100 μs, 9 nA, corresponding to 1.2 times the threshold value (lower traces)]. In both cases, the current injection was made at the soma. This is the same cell as shown in the upper panel of (A). (C) Summary graph of the swapping of action potential initiation from preferred to nonpreferred sites. The plot shows the time difference between the peaks of action potentials recorded simultaneously at two different locations for long low- intensity pulses (100 ms, 100 pA, applied to the preferred site) and brief high-intensity pulses (100 μs, 9 nA, corresponding to 1.2 to 2.3 times the threshold stimulus intensity with the same pulse duration, applied to the nonpreferred site). Data from 16 simultaneous somatic and dendritic recordings and from 3 simultaneous recordings from opposite dendrites at 22° to 25°C. Red circles represent cells in which initiation shifted toward the dendritic site; black triangles represent cells in which initiation shifted toward the somatic site. Measurements from the same cell are connected by lines. Qualitatively similar results were obtained at 33° to 36°C (not shown).

Once initiated, action potentials propagate with high velocity and constant amplitude over the somatodendritic membrane of oriens-alveus interneurons (Fig. 4). The ratio between action potential amplitudes at the dendrite and the soma was close to unity (0.97 ± 0.01; 27 cells), almost independent of distance (Fig. 4B, open symbols) (20). The mean conduction velocity, determined from the distance between recording sites and the time difference between the peaks of the action potentials evoked by threshold stimuli, was 0.91 ± 0.13 m s–1 (Fig. 4C, open bar). To assess the contribution of dendritic Na+channels to the nondecremental and fast spike propagation, we used action potentials as somatic voltage-clamp commands and recorded the resulting voltage changes at the dendrite (Fig. 4A, lower traces). In the presence of 500 nM TTX, the amplitude of the dendritic voltage signal was markedly attenuated (Fig. 4B, solid circles), and the conduction velocity was reduced to 0.24 ± 0.03 m s−1 (Fig. 4C, solid bar). Thus, voltage-activated Na+ channels mediate the active propagation of action potentials in interneuron dendrites. In various types of neurons, the safety factor of somatodendritic conduction is reduced during high-frequency stimulation (21). In contrast, in oriens-alveus interneurons, somatodendritic propagation was reliable for single action potentials and high-frequency trains (Fig. 4, D and E). The amplitude ratio of dendritic and somatic action potentials for the last spike evoked by 100-ms depolarizing current pulses was 0.97 ± 0.01 (27 cells), very similar to that for the first spike (P > 0.5) (Fig. 4E). These results show that somatodendritic propagation of action potentials in oriens-alveus interneurons is nondecremental, fast, and reliable, even during high-frequency stimulation. Thus, dendritic propagation in interneurons differs from that in cortical principal cells, where action potentials propagate with marked attenuation, low velocity (0.1 to 0.24 m s−1), and reduced reliability during high-frequency trains (17, 21).

Figure 4

Active somatodendritic propagation of action potentials and boosting of EPSPs. (A) Active and passive action potential propagation in the same oriens-alveus interneuron. Upper traces, dendritic (red) and somatic (black) action potential evoked by a current pulse (40 ms, 200 pA applied at the soma); lower traces, dendritic response evoked by the previously recorded somatic action potential used as a voltage-clamp (VC) command in the presence of 500 nM TTX. (B) Ratio of dendritic and somatic action potential (AP) amplitudes plotted versus the distance from the soma for the axon-bearing dendrite (open circles and continuous line at 22° to 25°C; triangles at 33° to 36°C) and the axon-lacking dendrite (squares and dashed line at 22° to 25°C; inverted triangles at 33° to 36°C). In the control, the ratio is close to unity, almost independent of location. In TTX, the ratio is smaller than unity and decreases with distance (solid circles and lower continuous line at 22° to 25°C). Points at a distance of 0 μm represent double-soma recordings. Mean somatic action potential amplitude is 115 ± 2 mV at 22° to 25°C and 98 ± 3 mV at 33° to 36°C. (C) Conduction velocities in active (open bar, 21 cells) and passive dendrites (solid bar, 4 cells), determined from the time difference between the peaks of action potentials at two sites or from the time difference between the peaks of the somatic voltage-clamp command and the dendritic response, respectively. For the active conduction velocity, a subset of cells in which the axon originated outside the two recording sites was analyzed. Error bars represent SEM. (D) Somatodendritic action potential propagation during a spike train. Lower traces are expanded versions of the first and the last action potential in the train, recorded simultaneously at the dendrite and the soma (100-ms, 100-pA pulse applied to the soma). (E) Ratio of dendritic and somatic action potential amplitudes for the last action potential in a train evoked by a 100-ms pulse, plotted against the distance of the dendritic recording site from the soma. Symbol codes are the same as in (B). Mean amplitude of last somatic action potential is 105 ± 2 mV at 22° to 25°C and 90 ± 3 mV at 33° to 36°C. For the first (B) and last (E) action potentials, amplitude ratios at low and high temperature were not significantly different (P > 0.1 to 0.9). (F) Na+ channel–mediated boosting of EPSPs. Artificial EPSPs recorded at the soma of an oriens-alveus interneuron, evoked by the injection of a dendritic current waveform with a rise time constant of 0.1 ms and a decay time constant of 1 ms. The peak current was increased from 0.133 to 2 nA in 0.133-nA steps. Upper traces were recorded in control conditions (action potentials in the last traces are truncated), and lower traces were recorded in the presence of 500 nM TTX in the bath.

Dendritic Na+ channels may boost synaptic events generated at distal dendritic locations (22). To test this hypothesis, we injected currents with time courses similar to those of excitatory postsynaptic currents (artificial EPSCs) into the dendrite, and recorded the corresponding voltage responses [artificial excitatory postsynaptic potentials (EPSPs)] at the soma (Fig. 4F). In control conditions, the relation between the EPSP and EPSC peak amplitudes was supralinear (seven cells). In contrast, in the presence of 500 nM TTX, the EPSP-EPSC relation was linear (four neurons), indicating that Na+ channels boost distal synaptic inputs (Fig. 4F). Boosting was slightly enhanced during trains of artificial EPSCs; for 25-ms interpulse intervals, the amplitudes of EPSPs in control conditions relative to those in TTX were 110 ± 3% for the first and 124 ± 4% for the third EPSP (six cells) (amplitude of the first EPSP was 13.6 ± 1.2 mV).

In conclusion, we have shown that voltage-gated channels in dendrites of oriens-alveus interneurons mediate dendritic action potential initiation, active spike propagation, and boosting of distal EPSPs. Dendritic action potential initiation ensures fast and reliable activation of oriens-alveus interneurons, which may be critical for the control of efficacy and plasticity of entorhinal inputs onto CA1 pyramidal neurons (8). Actively propagated dendritic action potentials could be important for the induction of associative long-term changes in the efficacy of glutamatergic synapses on oriens-alveus interneurons (23). Alternatively, actively propagated dendritic spikes may trigger GABA release from interneuron dendrites (24) or enhance dendrodendritic electrical coupling in interneuron networks (25). Boosting apparently contributes to the marked paired-pulse facilitation of EPSPs of excitatory synapses on oriens-alveus interneurons (26). Whether the presence of active dendritic conductances is a general property of cortical interneurons, including those mediating perisomatic inhibition, remains to be addressed.

  • * To whom correspondence should be addressed. E-mail: jonasp{at}


View Abstract

Navigate This Article