Activity-dependent downregulation of D-type K⁺ channel subunit Kv1.2 in rat hippocampal CA3 pyramidal neurons

Abstract

The intrinsic excitability of neurons plays a critical role in the encoding of memory at Hebbian synapses and in the coupling of synaptic inputs to spike generation. It has not been studied whether somatic firing at a physiologically relevant frequency can induce intrinsic plasticity in hippocampal CA3 pyramidal cells (CA3-PCs). Here, we show that a conditioning train of 20 action potentials (APs) at 10 Hz causes a persistent reduction in the input conductance and an acceleration of the AP onset time in CA3-PCs, but not in CA1-PCs. Induction of such long-term potentiation of intrinsic excitability (LTP-IE) was accompanied by a reduction in the D-type K⁺ current, and was abolished by the inhibition of endocytosis or protein tyrosine kinase (PTK). Consistently, the CA3-PCs from Kv1.2 knock-out mice displayed no LTP-IE with the same conditioning. Furthermore, the induction of LTP-IE depended on the back-propagating APs (bAPs) and intact distal apical dendrites. These results indicate that LTP-IE is mediated by the internalization of Kv1.2 channels from the distal regions of apical dendrites, which is triggered by bAP-induced dendritic Ca²⁺ signalling and the consequent activation of PTK.

Key points

• The intrinsic excitability of a hippocampal CA3 pyramidal cell (CA3-PC), but not CA1-PC, is enhanced by repetitive somatic firing at a physiologically relevant frequency (10 Hz for 2 s).

• Such an excitability change is mediated by the Ca²⁺- and Src family kinase-dependent endocytosis of D-type K⁺ channel subunit Kv1.2.

• We provide evidence that the surface expression of D-type K⁺ channels is higher in the distal apical dendrites than in the proximal apical dendrites in CA3-PCs.

• These results help us understand neuronal computational mechanisms underlying the cognitive functions of the hippocampal CA3 area.

Introduction

Long-term changes in the excitability of hippocampal CA3 pyramidal cells (CA3-PCs) is of crucial importance for understanding neuronal computational mechanisms underlying the proposed cognitive functions of the hippocampal CA3 area (O’Reilly & McClelland, 1994). Previously, it has been shown that intrinsic excitability of CA3-PCs can be altered by the chronic suppression of network activity (Cudmore et al. 2010), activation of metabotropic glutamate receptors (Brown et al. 2011) or intense somatic stimulation using a long depolarization and theta burst firing (Brown & Randall, 2009). However, it has not been investigated whether physiologically relevant repetitive somatic firing can induce intrinsic plasticity in CA3-PCs. Furthermore, input conductance (G_in) around the resting membrane potential (RMP) has profound effects on the threshold for somatic and dendritic spike generation in a neuron, but the activity-dependent change of G_in around the RMP has not been previously reported. In vivo recordings in the rat hippocampal CA3 pyramidal layer have shown that PCs in the CA3 field discharge at a frequency between 10 and 20 Hz in the place field (Leutgeb et al. 2007; McHugh et al. 2007). We inquired whether any changes in G_in around the RMP in the CA3-PC can be induced by somatic firing at the frequency that is found in the in vivo recordings. Through examination of the ion channel responsible for G_in changes induced by the somatic firing of CA3-PCs at moderate frequencies, we show that the D-type K⁺ current (I_K(D)) undergoes downregulation depending on the dendritic [Ca²⁺] elevation.

The D-type K⁺ current (I_K(D)) is a low-threshold, slowly inactivating and dendrotoxin-sensitive outward K⁺ current. I_K(D) has been characterized in hippocampal CA1 pyramidal cells (CA1-PCs; Storm, 1988) and in CA3-PCs (Lüthi et al. 1996). The activation and inactivation curves of I_K(D) overlap around the RMP resulting in a window current (Storm, 1988), which renders I_K(D) suitable for regulating the threshold for the generation of dendritic and somatic spikes (Golding et al. 1999; Cudmore et al. 2010). In addition, the rapid activation and slow inactivation properties of I_K(D) contribute to spike repolarization (Mitterdorfer & Bean, 2002) and to a delay and variability in the onset of the first action potential (AP; Cudmore et al. 2010), respectively.

I_K(D) in central neurons is primarily mediated by members of the Kv1 channel subfamily, Kv1.1 and Kv1.2. Although Kv1 channels are targeted to the axon (Wang et al. 1993; Gu et al. 2003), their dendritic expression is evident too in cortical pyramidal neurons (Sheng et al. 1994; Guan et al. 2006). Previous studies using a heterologous expression system have revealed that the internalization of the Kv1.2 channels is regulated by the Ca²⁺-dependent non-receptor tyrosine kinase PYK2 and its downstream Src-family kinases (Lev et al. 1995; Nesti et al. 2004). Although homeostatic regulation and Ca²⁺-dependent shift of the inactivation curve of I_K(D) has been reported (Saviane et al. 2003; Cudmore et al. 2010), activity-dependent modulation of I_K(D) has not been demonstrated in the context of neuronal intrinsic plasticity.

Here, we show that the excitability of CA3-PCs can be enhanced over the long term by repetitive somatic firing at a physiologically relevant firing frequency, and provide evidence that such changes in intrinsic excitability is mediated by Ca²⁺- and Src-family protein tyrosine kinase (PTK)-dependent internalization of the D-type K⁺ (K_D) channel subunits K_V1.2 from the distal apical dendrites.

Methods

Slice preparation

Hippocampal brain slices were prepared from Sprague–Dawley rats (P13–P18; P, postnatal days) or mice (P13–P18) of either sex as described previously (Lee et al. 2007). Protocols were approved by the Animal Care Committee of Seoul National University. Rats or mice were anaesthetized with isoflurane and decapitated. Brains were quickly removed and chilled in an ice-cold high-magnesium cutting solution containing the following (in mm): 116 NaCl, 26 NaHCO₃, 3.2 KCl, 0.5 CaCl₂, 7 MgCl₂, 1.25 NaH₂PO₄, 10 glucose, 2 sodium pyruvate, 3 ascorbate, with pH 7.4 adjusted by saturating with carbogen (95% O₂, 5% CO₂), and with osmolarity of ∼300 mosmol l⁻¹. The isolated brain was glued onto the stage of a vibratome (Leica VT1200) and 300 μm-thick transverse hippocampal slices were cut. The slices were incubated at 34°C for 30 min in the same solution, and thereafter maintained at room temperature until required. For experiments, we transferred a slice that had recovered for at least an hour to a recording chamber superfused with artificial cerebrospinal fluid (ACSF) containing (in mm): 124 NaCl, 26 NaHCO₃, 3.2 KCl, 2.5 CaCl₂, 1.3 MgCl₂, 1.25 NaH₂PO₄, 10 glucose, bubbled with 95% O₂ and 5% CO₂. All experiments were performed using the rat hippocampal slices except for Fig. 4B and C, where hippocampal slices from mutant mice or their wild-type littermates were used.

Figure 4. Downregulation of Kv1.2 underlies the LTP-IE induction.

Aa, DTX-I (black filled circles), but not DTX-K (open circles), occluded the effect of the conditioning AP train on G_in. The changes in G_in induced by the conditioning AP train (arrowhead) are indicated by black (DTX-I) and grey (DTX-K) arrows on the right of the graph. The control time course of G_in was reproduced from Fig. 1G (grey dots). Ab, summary for G_in, AP onset time, and the first-spike latency before and after the conditioning AP train in the presence of either DTX-I or DTX-K. B, the CA3-PCs from Kv1.2-KO (kcna2^−/−) mice exhibited significantly smaller I_K(D) than the wild-type (WT) mice. Ca, the conditioning AP train (10 Hz for 2 s) induced LTP-IE in the CA3-PCs of WT mice (open circles), but not in the Kv1.2-KO mice (filled circles). Cb, summary for the changes in G_in, AP onset time, and the first-spike latency induced by the conditioning AP train in the WT and the KO mice. Mean ± SEM; *P < 0.05; **P < 0.01; ***P < 0.005.

Electrophysiological recordings

All recordings were performed in the presence of antagonists of GABA_A and AMPA/kainite receptors (100 μm picrotoxin (PTX) and 10 μm 6-cyano-7- nitroquinoxaline-2,3-dione (CNQX), respectively). Whole-cell voltage- or current-clamp recordings from hippocampal CA3-PCs (one cell per slice) were carried out at 32 ± 2°C while the recording chamber was perfused with ACSF at 1–1.5 ml min⁻¹. The recordings were made using a MultiClamp 700B amplifier controlled by Clampex 10.2 via a Digidata 1440A data acquisition system (Molecular Devices, Sunnyvale, CA, USA) with a pipette solution containing (in mm): 130 potassium gluconate, 7 KCl, 2 NaCl, 1 MgCl₂, 0.1 EGTA, 2 ATP-Mg, 0.3 Na-GTP, 10 Hepes (pH 7.30 with KOH, 295 mosmol l⁻¹ with sucrose). After forming a whole-cell patch on the soma of a CA3-PC, the membrane potential was maintained at −67.1 ± 1.1 mV under current-clamp modes. Under this condition, we monitored input conductance (G_in) every 10 s before and after repetitive somatic stimulation with a short train of suprathreshold depolarizing current pulses (3 ms/900 pA pulses for 50 Hz train; 5 ms/700 pA pulses for other frequency trains). G_in was estimated from subthreshold voltage responses to −30 pA and +10 pA current steps (duration, 500 ms). The AP onset time and the first-spike latency are defined as the delay from the start of square and ramp depolarizing current injection to the beginning of the upstroke phase of the 1st evoked AP, respectively (Fig. 1D). In some experiments (Fig. 6), a deep cut was made at stratum oriens or at outer stratum radiatum about 250 μm from stratum lucidum for truncation of a part of the CA3-PC dendrites using a modified 20 G needle propelled by a micromanipulator.

Figure 1. Induction of long-term potentiation of intrinsic plasticity (LTP-IE) by somatic firing at different frequencies in CA3 pyramidal cells (CA3-PCs).

A, reduction in input conductance (G_in) after repetitive somatic firing at different frequencies (5, 10, 20 or 50 Hz for 2 s, arrowhead). Somatic firing was evoked by supra-threshold current injection via whole-cell patch pipette. G_in values were normalized to the baseline value. B, representative voltage responses to sub-threshold current injection (+10 and −30 pA) before (black) and 20 min after (red) the conditioning at different frequencies. C, summary for the relative G_in changes before and 20 min after the conditioning AP train at different frequencies in CA3-PCs. The ‘no stim’ in the left bar graph denotes the variance of the baseline G_in during the same recording time span. D, representative AP traces evoked by a step pulse (left, 400–500 pA, 9–13 ms) or a ramp current (right, 250 pA s⁻¹ for 1 s) before (black) and after (red) conditioning in CA3-PCs. Definitions of the AP onset time and the first-spike latency are depicted. E, summary for the AP onset time and the first-spike latency before and 20 min after the conditioning AP train at different frequencies in CA3-PCs. F, somatic 10 Hz AP train-induced reduction of G_in in longer time span (up to 50 min). Without the conditioning, G_in was not significantly altered (filled circles). Inset, exemplar voltage traces before (black) and 13 min, 33 min and 45 min after the conditioning (red). G, relative G_in changes in CA1-PCs (grey open circles) and CA3-PCs (black open circles) after the conditioning (somatic firing at 10 Hz for 2 s, arrow). Inset, exemplar subthreshold voltage responses before (black) and 20 min after (red) the conditioning in CA1- (upper) and CA3-PCs (lower). For CA1-PCs, smaller current pulses (+3 and −5 pA) were injected to avoid a voltage sagging. H, summary for the relative G_in changes, the AP onset time and the first-spike latency before and 20 min after the conditioning AP train in CA3- and CA1-PCs. Mean ± SEM. Error bars, SEM. The first-spike latency was normalized to the baseline because of large cell-to-cell variation.

Figure 6. The induction of LTP-IE involves distal dendritic events.

Aa, LTP-IE was not induced in the CA3-PCs whose distal apical dendrites were truncated (open triangles). Inset shows representative subthreshold voltage responses before (black) and after (red) the conditioning in the truncated neuron. Ab, the baseline G_in in the truncated neurons. Ac, representative AP responses before and after the conditioning in the truncated neurons. The truncation was performed at the site 250 μm from the soma (differential interference contrast (DIC) images). The right image shows the truncated region of the left image. Ad, summary of excitability parameters in truncated neurons. Ba and Bb, representative traces of low 4-AP-sensitive current (I_K(D)) under control conditions, after truncation of distal apical dendrites (ca 250 μm from the soma, left DIC image of Bb), and after a cut was made at stratum oriens (right image, scale bar, 300 μm). The mean values for the peak I_K(D) are summarized in the bar graph of Bb.

D-type K⁺ current recordings

Somatic K⁺ outward current was recorded in the voltage-clamp mode in the presence of synaptic blockers (PTX and CNQX) plus an inward current blocker cocktail that consisted of 200 μm Ni²⁺, 125 μm Cd²⁺ and 0.5 μm tetrodotoxin (TTX) in order to block voltage-dependent Ca²⁺ and fast Na⁺ channels. The D-type K⁺ current (I_K(D)) was isolated by algebraic subtraction of the K⁺ outward current in the presence of 30 μm 4-aminopyridine (low 4-AP) from that in the control conditions. The peak current amplitude of I_K(D) in a cell was measured from the averaged trace of 5–10 D-type K⁺ current traces.

Kv1.2 knock-out mice

Heterozygous Kv1.2 knock-out mice (kcna2^+/−) were purchased from The Jackson Laboratory (Bar Harbor, ME, USA; Donating Investigator: Dr Bruce Tempel, Univ. of Washington School of Medicine), and maintained at the approved specific-pathogen-free facilities. By inter-crossing kcna2^+/− mice, we bred homozygous knock-out mice (kcna2^−/−) and wild-type littermates (kcna2^+/+). For genotyping, DNA was isolated from the tail of each mouse in a litter aged 6–8 days as described by Brew et al. (2007). Detailed protocols are available online (http://depts.washington.edu/tempelab/Protocols/KCNA2.html). Although kcna2^−/− mice had a severely reduced life span (range of P18–P23; Brew et al. 2007), they appeared normal during the first 2 weeks of their life.

Hippocampal organotypic slice culture

Organotypic slice cultures were prepared from P6–P8 Sprague–Dawley rat hippocampus as described previously (Lee et al. 2012) according to the protocol approved by the Animal Care Committee of Seoul National University. Briefly, hippocampal slices (thickness, 300 μm) were obtained using a vibratome (DTK-1000 ZERO 1; Dosaka) and placed on a porous (0.4 μm) membrane (Millicell-CM; Millipore). The culture medium was a mixture of 50% minimum essential medium (Invitrogen, Carlsbad, CA, USA), 25% horse serum, 24% Eagle's buffered salt solution, and 1% penicillin–streptomycin (PS, Invitrogen). Glucose was added to reach a final concentration of 36 mm. After 3 days in culture, the medium was changed to serum-free Neurobasal medium with 2% B-27 supplement, 1% GlutaMAX-I, 1% PS and 5 mm glucose added. To arrest glial proliferation, 5 μm Ara-C was also added to the culture medium and the medium was changed every 2 days.

Gene knock-down using RNA interference

We used a short-hairpin RNA (shRNA) to deplete endogenous PYK2 from the hippocampal CA3 pyramidal neuron in an organotypic slice culture. PYK2-targeting short-interfering (si) RNA sequences (5′-GCTGTAGCATAGAGTCAGA-3′) was referred to Hsin et al. (2010). The luciferase-targeting siRNA sequence (5′-TAAGGCTATGAAGAGATAC-3′) was used as a non-targeting siRNA control (shNT). The synthesized PYK2-targeting or non-targeting siRNA oligonucleotides (Cosmogenetech, Seoul, Korea) were ligated into the lentiviral vector pLentiLox3.7 (pLL3.7), which co-expresses green fluorescent protein (GFP). To confirm the knock-down effect of PYK2-targeting shRNA (shPYK2), primary cultured hippocampal neurons (DIV16) were infected with lentivirus encoding shPYK2 or non-targeting shRNA (shNT). On the 5th day after infection, neurons (DIV21) were lysed in an ice-cold lysis buffer containing 50 mm Tris (pH 7.4 adjusted with HCl), 150 mm NaCl, 1 mm EDTA and 1% SDS, and the lysate was subject to immunoblotting analysis. The procedures for hippocampal slice culture, lentivirus production and immunoblotting were previously described in detail in Lee et al. (2012). Primary and secondary antibodies were used as follows: rabbit monoclonal anti-PYK2 (1:1000; Abcam, Cambridge, MA, USA), mouse monoclonal anti-β-actin (1:5000; Santa Cruz Biotechnology, Santa Cruz, CA, USA), horseradish peroxidase (HRP)-conjugated goat anti-rabbit IgG (1:2000; Abcam) or HRP-conjugated donkey anti-mouse IgG (1:5000; Jackson ImmunoResearch, West Grove, PA, USA). CA3-PCs in the cultured slices at DIV3–5 were transfected with plasmids encoding shPYK2/GFP or shNT/GFP using the biolistic transfection methods (Helios Gene Gun System, BioRad). The transfected CA3-PCs were subject to electrophysiological recording 5 days after the transfection.

Cytosolic Ca²⁺ measurements

The procedures for cytosolic [Ca²⁺] measurement in the slice have been previously described in detail (Lee et al. 2009). Briefly, we measured cytosolic [Ca²⁺] from fluorescence images of a hippocampal CA3-PC in the slices loaded with fura-2 (pentapotassium salt, 0.1 mm) via a whole-cell patch pipette. Imaging of a CA3-PC was performed with a ×60 water immersion objective (NA 0.9, LUMPlanFI, Olympus, Japan), an air-cooled slow scan CCD camera (SensiCam, PCO, Germany) and a monochromator (xenon-lamp based, Polychrome-IV, TILL-Photonics, Martinsried, Germany), which were controlled by a PC and Digidata 1440A, running a custom-made software programmed with Microsoft Visual C++ (version 6.0; CCDLabo1.2). To increase the time resolution and minimizing the photobleaching effect, we adopted the single-wavelength protocol in the Ca²⁺-imaging of the slice. Images taken at 20 Hz with single wavelength excitation at 380 nm (F₃₈₀) were preceded and followed by images with excitation at isosbestic wavelength (360 nm). The isosbestic fluorescence values (F_iso) were linearly interpolated between points just before and after the period of excitation at 380 nm. Calibration parameters were determined using in cell calibration. The ratio (R) of F_iso/F₃₈₀ was converted to [Ca²⁺] values using equation: [Ca²⁺]=K_eff(R–R_min)/(R_max–R_min), where R_min and R_max are R values at zero and 10 mm[Ca²⁺]_i. K_eff was determined using a pipette solution of known [Ca²⁺], which contained 10 mm BAPTA and 10 mm CaCl₂.

Data analysis

Data were analysed using IgorPro (version 6.10A; WaveMetrics, Lake Oswego, OR, USA). Statistical data are expressed as mean ± standard error of the mean (SEM), and n indicates the number of cells studied. The significance of differences between two experimental conditions was evaluated using the non-parametric Mann–Whitney U test and Wilcoxon signed rank test for non-paired and paired data, respectively, using a significance level (P) of 0.05.

Results

We studied the long-term changes in input conductance (G_in) in CA3-PCs induced by their repetitive somatic firing at different frequencies (5, 10, 20 or 50 Hz) under the current clamp mode of whole-cell patch clamp techniques (see online Supplemental Results and Supplemental Fig. 1 for representative membrane potential responses to the induction protocols). G_in was determined every 10 s from the voltage deflections elicited by positive and negative current injections (+10 and −30 pA, respectively) at the RMP (−67.1 ± 1.1 mV; exemplar voltage traces, Fig. 1B). In addition to G_in, we examined the effect of repetitive somatic firing on the delay to the onset of the first spike elicited by the injection of a short depolarizing current step (400–500 pA for 9–13 ms) or a ramp current (250 pA s⁻¹ for 1 s; Fig. 1D). Henceforth, the former and latter are referred to as the AP onset time and the first-spike latency, respectively. Statistical values and significance for the effects of the conditioning on G_in, AP onset time and first-spike latency were determined 20 min after the conditioning, and are summarized online in Supplemental Table 1.

Repetitive somatic firing at an optimal frequency induces a long-term reduction of G_in in CA3-PCs

After assessing the baseline G_in, we delivered a train of suprathreshold current pulses for 2 s at different frequencies, and then resumed the monitoring of G_in (Fig. 1A). Although G_in was not altered by somatic firing at 5 Hz (98.6 ± 1.5%, P= 0.56, n= 6), it underwent a gradual and persistent reduction at 10 Hz (67.8 ± 1.6%, P < 0.005, n= 19). The G_in reduction at 20 Hz was less pronounced than at 10 Hz (80.4 ± 1.9%, P < 0.005, n= 11), and was negligible at 50 Hz (95.5 ± 3.7%, P= 0.22, n= 13; Fig. 1A), indicating that there is an optimal somatic firing frequency that induces the reduction of G_in in CA3-PCs. Figure 1C shows the relative changes in G_in measured 20 min after the conditioning as a function of the somatic firing frequency. This form of intrinsic plasticity lasted as long as we held the whole-cell recordings (up to 50 min), but no significant change in G_in was observed in the naive CA3-PCs (Fig. 1F).

The change in G_in was accompanied by decreases in the AP onset time and first-spike latency (Fig. 1D and E). Figure 1E summarizes the mean values for the AP onset time and relative changes in the first-spike latency after conditioning with somatic firing at different frequencies, showing that somatic firing at 10 Hz for 2 s has the greatest effect on both parameters. Henceforth, we refer to somatic firing at 10 Hz for 2 s as the ‘conditioning AP train’, and to long-term enhancement of intrinsic excitability induced by the conditioning AP train as ‘long-term potentiation of intrinsic excitability’ or ‘LTP-IE’. It should be noted that the term ‘LTP-IE’ has been used to represent a variety of intrinsic plasticities with different characteristics in different brain regions (Sourdet et al. 2003; Cudmore & Turrigiano, 2004; Xu et al. 2005).

In contrast to the G_in of CA3-PCs, G_in of CA1-PCs was not altered by the same conditioning AP train (104.3 ± 3.7%, P= 0.31, n= 6; Fig. 1G). Figure 1H summarizes the changes in the mean values for G_in, AP onset time and relative changes in the first-spike latency after the conditioning AP train in CA3-PCs (control, n= 19, P < 0.005 for all three parameters) and CA1-PCs (n= 8, P > 0.1 for all). These findings indicate that the intrinsic excitability of CA3-PCs, but not of CA1-PCs, is enhanced by their somatic firing at the physiologically relevant frequency, and that this form of LTP-IE is not a general property of pyramidal neurons.

The induction of LTP-IE in CA3-PCs requires an elevation of intracellular [Ca²⁺]

Ca²⁺ plays a crucial role in inducing various forms of activity-dependent intrinsic plasticity (Cudmore & Turrigiano, 2004; Frick et al. 2004; Fan et al. 2005). To test whether the induction of LTP-IE requires intracellular [Ca²⁺] elevation, we included BAPTA (10 mm), a Ca²⁺ chelator, in the pipette solution, and repeated the same experiments as in Fig. 1G. Intracellular BAPTA or bath-applied nimodipine (10 μm), an L-type voltage-dependent Ca²⁺ channel (VDCC) blocker, abolished the conditioning AP train-induced reduction of G_in (BAPTA, 110.7 ± 3.2%, n= 15; nimodipine, 107.7 ± 5.7%, n= 8, Fig. 2A). Similarly, BAPTA or nimodipine abolished the reduction of the AP onset time and first-spike latency that are normally observed after the conditioning (Fig. 2B and C). These results suggest that the induction of LTP-IE is triggered by Ca²⁺ coming through L-type VDCCs which are activated by the conditioning AP train.

Figure 2. The LTP-IE induction depends on Ca²⁺ signalling.

A and B, intracellular BAPTA (light red circles, 10 mm) or bath-applied nimodipine (red circles, 10 μm) abolished the conditioning AP train-induced reduction of G_in (A), AP onset time and the first-spike latency (B) in the CA3-PCs. Each superimposed pair of traces show exemplar voltage traces before (black) and 20 min after (red) the conditioning. C, summary for relative changes in G_in, AP onset time, and relative changes in the first-spike latency before and after the conditioning AP train in the CA3-PCs under control conditions and in the presence of BAPTA or nimodipine. Mean ± SEM.

Next, we tested the possible involvement of NMDA receptors in the induction of LTP-IE. Because the CA3-PC was conditioned by somatic firing without stimulation of afferent fibres, it is unlikely that NMDA receptors subserve the Ca²⁺ influx required for the induction of LTP-IE. Supporting this idea, bath application of 40 μm MK-801 had no significant effect on the induction of LTP-IE by the same conditioning (see Supplemental Fig. 2A). d-Aminophosphonovalerate (APV), another NMDA receptor blocker, was not appropriate for testing the involvement of NMDA receptors in LTP-IE because it reduced the baseline G_in and the first-spike latency (Supplemental Fig. 2B).

LTP-IE is mediated by the downregulation of D-type K⁺ channels

The change in G_in around RMP implies that the induction of LTP-IE involves an alteration of ion channels that are active near the RMP. We found that LTP-IE persists in the presence of ZD7288, apamin or linopirdine, which blocks hyperpolarization-activated cation (HCN), small-conductance Ca²⁺-activated K⁺ (SK) and Kv7 channels, respectively (Supplemental Fig. 3). The Kv1 family current or D-type K⁺ current (I_K(D)) is activated below the AP threshold (Storm, 1988), and its magnitude has profound effects on the input resistance near the RMP (Saviane et al. 2003). Furthermore, the slow inactivating kinetics of I_K(D) causes a delayed onset of AP upon injection of a depolarizing current step (Cudmore et al. 2010). The finding that the conditioning AP train results in both a reduction of G_in and an acceleration of the AP onset time prompted us to test the possible involvement of I_K(D) in the LTP-IE induction.

It is well known that a low concentration of 4-aminopyridine (4-AP, 30–50 μm) specifically blocks I_K(D) (Storm, 1988). Bath application of 30 μm 4-AP caused a significant decrease in the baseline G_in (80.2 ± 1.7%, P < 0.05, n= 7; Fig. 3Aa). In the presence of low 4-AP, the conditioning AP train instead slightly increased G_in (86.8 ± 3.3%, Fig. 3Aa). Furthermore, once G_in was reduced by the conditioning, 30 μm 4-AP did not cause a further reduction in G_in (71.1 ± 1.8% to 70.5 ± 2.3%, P= 0.93, n= 7; Fig. 3Ba). Low concentration 4-AP also occluded the conditioning-induced reduction of the AP onset time and first-spike latency. Both parameters were significantly reduced by low 4-AP (P < 0.05 for both, n= 5; Fig. 3Ab). In the presence of 4-AP, the conditioning AP train had no additional effect on either the AP onset time or first-spike latency (P > 0.05 for both, Fig. 3Ab). Reciprocally, the conditioning occluded the effect of 4-AP on the AP onset time and first-spike latency (P > 0.05 for both, n= 5, Fig. 3Bb).

Figure 3. The LTP-IE induction is mediated by downregulation of D-type K⁺ current.

A and B, the effect of the conditioning AP train was occluded by low concentration 4-AP (30 μm) and vice versa in the CA3-PCs. Aa, time course of relative G_in change during bath application of low 4-AP, which was followed by the conditioning AP train (arrow). Inset shows representative sub-threshold voltage responses. Ba, G_in changes caused by the conditioning AP train (arrow) and subsequent bath-application of low 4-AP. Ab and Bb, summary of the effects of the conditioning and low 4-AP on G_in, the AP onset time, and the first-spike latency. C, the D-type K⁺ current (I_K(D)) was reduced in the CA3-PCs by the conditioning AP train. Ca and Cb, outward currents caused by a depolarizing step to −20 mV from −70 mV before and after low 4-AP treatment in the naive and conditioned CA3-PCs. For clarity, points on the fast capacitive discharge were removed from raw traces. Below the raw traces, 4-AP-sensitive current traces obtained by subtraction of the latter from the former are shown and regarded as I_K(D). Cc, averaged traces for I_K(D) before and after the conditioning are superimposed (left). The conditioning AP train significantly reduced the peak amplitude of I_K(D) induced by step depolarization to −20 mV (right).

Next, we studied whether I_K(D) is reduced in CA3-PCs by the conditioning AP train. We regarded the slowly inactivating K⁺ outward current that is sensitive to 30 μm 4-AP as I_K(D) (Fig. 3C). To this end, K⁺ outward current needs to be isolated by bath application of a cocktail of inward current and synaptic current blockers (see Methods). Because conditioning is not possible in the presence of this cocktail, we could not study conditioning-induced changes of I_K(D) in the same cell. Instead, we compared the amplitude of I_K(D) at the test potential of −20 mV between the control and conditioned cell groups. In the naive CA3-PCs, the outward current was elicited by a depolarization step to −20 mV for 2 s in the presence of the blocker cocktail. The outward current exhibited fast- and slowly-inactivating and non-inactivating components (Fig. 3Ca). Additional bath application of 30 μm 4-AP abolished the slowly inactivating component. I_K(D) was dissected out by algebraic subtraction between the two current traces before and after 4-AP application (Fig. 3Ca). The I_K(D) that was measured using the same methods 20 min after the conditioning was significantly smaller than that under the control conditions (control, 359 ± 23 pA, n= 12; after conditioning, 125 ± 11 pA; P < 0.005, n= 15; Fig. 3C).

Downregulation of Kv1.2 underlies the induction of LTP-IE

The above results suggest that LTP-IE induction is mediated by the downregulation of D-type K⁺ (K_D) channels. Homo- or heterotetramer of Kv channel subunits constitutes a K_D channel. Kv1.1 and Kv1.2 are the two most abundant types of Kv1 subunits in the central neurons (Scott et al. 1994). To narrow down which Kv1 channel subunit is involved, we tested the effects of dendrotoxin-K (DTX-K, 50 nm) and dendrotoxin-I (DTX-I, 50 nm), which are selective blockers of Kv1.1- and Kv1.1/2/6-containing channels, respectively (Wang et al. 1999). The G_in measured 20 min after the conditioning was not different under the DTX-I or DTX-K conditions (P= 0.78, Fig. 4Aa). The conditioning-induced relative reduction of G_in in the presence of DTX-I, however, was significantly smaller compared to DTX-K (DTX-I, 11.6 ± 2.3; DTX-K, 26.6 ± 1.53%, P < 0.005, Fig. 4Aa and b), probably because the baseline G_in in the presence of DTX-I (2.9 ± 0.3 nS, n= 7), but not DTX-K (3.5 ± 0.4, n= 8), was slightly lower than the control value (3.5 ± 0.2 nS, n= 21, P= 0.13; Fig. 4Aa and Supplemental Table 1). This result implies that the conditioning-induced reduction of G_in is more efficiently occluded by DTX-I than DTX-K, and thus raises a possibility that LTP-IE is primarily mediated by downregulation of Kv1.2 and/or Kv1.6, which are sensitive to DTX-I but not to DTX-K. Accordingly, DTX-I, but not DTX-K, abolished the conditioning-induced reduction of the AP onset time and first-spike latency (n= 6; Fig. 4Ab).

Between the two Kv1 channel subunits blocked by DTX-I, Kv1.2 comprises the majority of the K_D channels (Shamotienko et al. 1997; Coleman et al. 1999). Furthermore, CA3-PCs exhibit a higher expression of Kv1.2 than CA1-PCs (Sheng et al. 1994). Therefore, we tested whether LTP-IE can be induced in the CA3-PCs of Kv1.2 knock-out (KO) mice (Brew et al. 2007). Prior to this experiment, we isolated I_K(D) using the same methods as in Fig. 3C for Kv1.2-KO mice and their wild-type (WT) littermates, and confirmed that the CA3-PCs in Kv1.2-KO mice have a significantly smaller I_K(D) (WT, 314 ± 29 pA, n= 7; KO, 149 ± 15 pA, n= 5; P < 0.005; Fig. 4B). In the CA3-PCs of WT mice (kcna2^+/+), the same conditioning AP train induced a gradual and persistent reduction of G_in (69.4 ± 2.3%; P= 0.02, n= 7; Fig. 4Ca). Both the AP onset time and first-spike latency were significantly reduced by the conditioning AP train too (P < 0.05, n= 5; Fig. 4Cb). The baseline G_in in the KO mice was already lower by ca 30% than that in the WT mice (WT, 3.6 ± 0.3 nS, n= 17; KO, 2.5 ± 0.4 nS, n= 5, P < 0.05), and was not further reduced by the conditioning (103.4 ± 5.2%, n= 5, Fig. 4Ca). Similarly, the AP onset time in the CA3-PCs of the KO mice was already faster than that in the WT mice (8.5 ± 0.6 ms vs. 5.9 ± 0.7 ms; P < 0.05; n= 5), and was not further accelerated by the conditioning (6.4 ± 0.6 ms, P= 0.38; Fig. 4Cb). The conditioning did not alter the first-spike latency in the KO mice either (P= 0.13, n= 5; Fig. 4Cb). These results indicate that the downregulation of Kv1.2 underlies the induction of LTP-IE.

Protein tyrosine kinase (PTK) is a downstream effector of Ca²⁺ for the induction of LTP-IE

Protein kinase A (PKA) has been suggested to mediate not only intrinsic plasticity in neocortical and CA1 pyramidal neurons (Hoffman & Johnston, 1998; Cudmore & Turrigiano, 2004; Hammond et al. 2008; Lin et al. 2008) but also the trafficking of Kv1.2 in the heterologous system (Connors et al. 2008). To study the possible involvement of PKA in LTP-IE, we examined the effect of KT-5720, a PKA inhibitor, on LTP-IE induction. In the presence of KT-5720 (1.2 μm), the conditioning AP train induced a robust decrease in G_in (71.4 ± 2.6%, P < 0.005, n= 10; Fig. 5A), and the acceleration of the AP onset time and first-spike latency (Fig. 5D), indicating that the LTP-IE induction is independent of the activation of PKA.

Figure 5. Activation of protein tyrosine kinase (PTK) and subsequent endocytosis of Kv1.2 underlie the LTP-IE induction.

A and B, the conditioning AP train-induced reduction of G_in was not affected by bath application of 1.2 μm KT-5720, but was abolished by bath or patch-pipette application of genistein (black open circles) or PP2 (black filled circles). C, dynasore (filled circles) included in the patch-pipette abolished the LTP-IE induction. The control time courses of G_in (grey circles) in A–C are reproduced from Fig. 1G. D, summary for the excitability parameters under different conditions shown in A–C. E, effects of PYK2-targeting shRNA (shPYK2) on the LTP-IE induction. The efficiency of shPYK2 was confirmed in cultured hippocampal neurons by western blot analysis of endogenous PYK2 (Supplemental Fig. 5). Ea, the conditioning AP train (arrowhead) induced a reduction of G_in in the CA3-PCs transfected with non-targeting shRNA (shNT), but not in the shPYK2-transfected CA3-PCs (shPYK2, 99.1 ± 3.38%, n= 12; shNT, 74.2 ± 3.67%, n= 9). CA3-PCs in the organotypic culture were biolistically transfected with plasmids encoding shNT or shPYK2 plus GFP (inset). Eb, summary of the effects of shNT and shPYK2 on excitability parameters. Neither AP onset time (P= 0.61) nor the first-spike latency (P= 0.20, n= 9) was altered by the conditioning in PYK2-depleted cells (P < 0.05 for both, n= 9). Note that the baseline AP onset time is significantly faster in the PYK2-depleted CA3-PCs (n= 10, P < 0.05). Mean ± SEM; *P < 0.05; **P < 0.01; ***P < 0.005.

The Kv1.2 channel activity is regulated by tyrosine phosphorylation involving proline-rich tyrosine kinase 2 (PYK2), a Ca²⁺-dependent non-receptor PTK (Lev et al. 1995). PYK2 can be stimulated by depolarization-induced Ca²⁺ influx (Lev et al. 1995), and its transcripts are abundant in the hippocampal CA3 region (http://mouse.brain-map.org). We tested the possibility that PYK2 is a downstream effector of the Ca²⁺ signalling that induces LTP-IE. Bath or patch-pipette application of genistein (100 μm), a broad-spectrum PTK inhibitor, abolished the conditioning-induced reduction of G_in (n= 6, P= 0.31, Fig. 5B), AP onset time and the first-spike latency (Fig. 5D). Although the activation of PYK2 is triggered by Ca²⁺, it is maintained by reciprocal phosphorylation between PYK2 and Src family kinases (SFKs; Girault et al. 1999; Corvol et al. 2005). Accordingly, PP2 (10 μm), a selective inhibitor of SFKs, abolished the conditioning-induced reduction of G_in (n= 7, Fig. 5B) and the AP onset time (Fig. 5D). These results suggest that PTK, but not PKA, is the downstream effector of Ca²⁺ which mediates the downregulation of Kv1.2.

It has been shown that endocytosis of Kv1.2 from the cell surface underlies the PTK-dependent suppression of Kv1.2 current (Nesti et al. 2004). We tested the effect of dynasore (40 μm), a dynamin inhibitor, on the induction of LTP-IE. When dynasore was included in the pipette solution, the conditioning AP train failed to reduce G_in (117.4 ± 5.1%; n= 8; Fig. 5C), the AP onset time and the first-spike latency (P= 0.75 and 0.63, respectively; n= 6; Fig. 5D), suggesting that Ca²⁺-dependent activation of PTK and endocytosis of Kv1.2 underlie the LTP-IE induction. Furthermore, the activation and inactivation curves of low 4AP-sensitive current were not altered by the conditioning AP train, supporting the essential role of channel trafficking in the activity-dependent regulation of Kv1.2 (Supplemental Fig. 4).

To test the possible involvement of PYK2 in the LTP-IE induction, we studied the conditioning-induced change of G_in in the CA3-PCs where PYK2 was depleted using shRNA targeting PYK2 (shPYK2; Hsin et al. 2010; see Supplemental Fig. 5 for the knock-down efficiency of shPYK2). Unexpectedly the baseline values for AP onset time and first-spike latency were faster in the shPYK2-transfected CA3-PCs than the shNT-transfected control, while the baseline G_in was not different between the two groups (Fig. 5Eb, Supplemental Table 1). Therefore, although the LTP-IE induction was abolished in the shPYK2-transfected CA3-PCs, we could not reach a conclusion on the involvement of PYK2 in the LTP-IE induction.

Surface expression of D-type K⁺ channels may be higher in the distal apical dendrites than in the proximal apical dendrites in CA3-PCs

The expression of ion channels is not homogeneous along the somatodendritic axis of a pyramidal neuron (Hoffman et al. 1997; Lorincz et al. 2002; Nusser, 2009). It has been shown that K_D channels are localized in axonal compartments including the axon initial segment (Gu et al. 2003; Goldberg et al. 2008). If the axonal K_D channels are involved in the LTP-IE induction, truncation of the distal apical dendrites is expected not to alter the LTP-IE induction. To test this hypothesis, we surgically truncated the distal part of the apical dendrites of CA3-PCs (ca 250 μm from the soma, see Methods), and tested whether LTP-IE can be induced by the conditioning in such CA3-PCs. The baseline G_in in the truncated neurons was already lower than that in the intact neurons (intact, 3.5 ± 0.2 nS, n= 21; after distal dendritic truncation, 2.4 ± 0.1 nS, n= 9, P= 0.014, Fig. 6Ab), and was not further reduced by the conditioning (104.6 ± 2.0%, n= 9, Fig. 6Aa). Similarly, the AP onset time and first-spike latency were already reduced in the truncated neurons compared to the intact ones (P < 0.05, respectively; n= 9; Fig. 6Ad), and was not further accelerated by the conditioning (Fig. 6Ac and Ad). These results argue against the involvement of the axonal K_D channels in LTP-IE, and rather suggest that LTP-IE involves the suppression of K_D channels expressed in the distal apical dendrites.

The reduction of G_in by truncation of the distal dendrites (Fig. 6Ab) implies a possible contribution of K_D channels expressed in the distal apical dendrites to the somatic recordings of I_K(D). To test this possibility, we dissected out the K⁺ outward current that is sensitive to 30 μm 4-AP at –20 mV in the CA3-PCs whose distal (>250 μm from the soma) apical dendrites were surgically truncated. The low 4-AP-sensitive outward current was significantly smaller in the truncated neurons (intact, 314 ± 29 pA, n= 10; after distal dendritic truncation, 85 ± 8 pA, n= 5, P < 0.005; Fig. 6B). To rule out any effect exerted by mechanical damages on the CA3-PCs, a cut was made on the stratum oriens where the basal dendrites of CA3-PCs exist. The cut on the stratum oriens reduced I_K(D) to a significantly smaller extent than the truncation of the distal dendrites (218 ± 14 pA, n= 5, P < 0.01, n= 5; Fig. 6Bb). Furthermore, the K⁺ outward current remaining after the treatment of 30 μm 4-AP was not significantly reduced by the truncation of the distal apical dendrites or basal dendrites, arguing against non-specific inhibition of ion channels due to the truncation of dendrites (Supplemental Fig. 6). These results suggest that K_D channels expressed in the distal region of apical dendrites contribute to the somatic recordings of I_K(D), and support the idea that K_D channels expressed in the distal dendrites rather than those in the soma or the axon are the major players in the LTP-IE.

Dendritic Ca²⁺ signalling is essential for the induction of LTP-IE

A somatic AP back-propagates into a dendritic arbor, and thus can induce Ca²⁺ entry through VDCCs in the dendritic compartments (Spruston et al. 1995; Kim et al. 2012). To test whether the LTP-IE induction requires somatic or dendritic Ca²⁺ signalling, we studied the dependence of the LTP-IE induction on back-propagating APs (bAPs). Tetrodotoxin (TTX) at low concentrations (10 nm) can selectively block bAPs with little effect on somatic APs in CA1-PCs (Mackenzie & Murphy, 1998; Magee & Carruth, 1999). To confirm this idea in CA3-PCs, we measured Ca²⁺ transients (CaTs) elicited by the conditioning AP train (10 Hz for 2 s) in the soma and in the distal apical dendritic region (200–250 μm from the soma) using calcium indicator dye fura-2 (0.1 mm). Bath-applied 10 nm TTX reduced the peak of the CaTs at the distal dendritic region to half of the control value (0.32 ± 0.04 μm to 0.15 ± 0.03 μm, n= 5), whereas it had little effect on the somatic CaTs (0.43 ± 0.02 to 0.41 ± 0.06 μm, n= 6; Fig. 7A), indicating that 10 nm TTX selectively suppresses the dendritic CaT induced by the conditioning AP train in the CA3-PCs. To determine whether somatic or dendritic Ca²⁺ signalling is required for the LTP-IE induction, we examined the effect of 10 nm TTX. The somatic firing induced by the conditioning AP train was not affected by 10 nm TTX (Fig. 7A). Nevertheless, the LTP-IE was abolished (97.7 ± 4.1%, n= 11; Fig. 7B and C) in the presence of 10 nm TTX, suggesting that bAPs and consequent dendritic Ca²⁺ signalling are crucial for the LTP-IE induction. This result further supports the idea that LTP-IE is associated with Ca²⁺-dependent downregulation of K_D channels expressed in the apical dendrites rather than in the soma or the axon.

Figure 7. Back-propagating AP is required for the induction of LTP-IE.

A, upper left: fluorescence images of the soma and distal dendrites (230–270 μm from the soma) of a CA3-PC filled with fura-2 via a somatic recording pipette. Scale bars, 30 μm. Upper right: representative Ca²⁺ transients (CaTs) measured from the soma (upper traces) and from distal dendrites (lower traces, the box in the dendritic fluorescence image) before (black) and after (blue) bath application of 10 nm tetrodotoxin (TTX). Lower left, somatic AP trains were little affected by 10 nm TTX (light blue). Lower right, summary of the effects of 10 nm TTX on the peak of somatic and dendritic CaTs. B and C, G_in was not reduced by the conditioning AP train (arrowhead) in the presence of 10 nm TTX (blue dots). Lower right, representative subthreshold voltage responses under control conditions (black), before (blue) and after (red) the conditioning in the presence of TTX.

Discussion

The key finding of the present study is that repetitive somatic firing of a CA3-PC at a moderate frequency induces PTK-dependent downregulation of Kv1.2 from distal apical dendrites. It is generally accepted that Kv1 channels are predominantly localized in axonal compartments (Monaghan et al. 2001; Gu et al. 2003). Accordingly, the role of I_K(D) has been investigated with regard to somatic or axonal AP generation (Saviane et al. 2003; Cudmore et al. 2010). Nevertheless, our results from the dendritic truncation experiments (Fig. 6), together with the effect of low concentration TTX (Fig. 7), indicate that K_D channels expressed in the apical dendrites are involved in the LTP-IE induction. Indeed, there is evidence for dendritic localization of K_D channel subunits (Sheng et al. 1994; Grosse et al. 2000), but there is no previous study on the role of I_K(D) in the activity-dependent regulation of dendritic excitability. The subcellular distribution of K_D channels along the somatodendritic axis is not known. Our truncation studies suggest that the surface expression of K_D channels over the dendritic arbor in a CA3-PC is not homogeneous but higher in the distal apical dendrites. However, it needs high resolution immuno-localization of the surface Kv1.2 subunits to elucidate their distribution along the apical dendrites in CA3-PCs.

Kv1.2 channel is a key player of LTP-IE in the CA3-PCs

In contrast to CA3-PCs, CA1-PCs exhibited no change in the intrinsic excitability by the same conditioning (Fig. 1G). Consistent with this finding, a previous report and in situ hybridization results available in the public domain indicate that CA3-PCs, among three types of hippocampal principal cells, exhibit the highest signal for the Kv1.2 message (Sheng et al. 1994; http://mouse.brain-map.org). Given that PTK-mediated internalization of Kv1.2 is responsible for LTP-IE in CA3-PCs, lower expression of Kv1.2 in CA1-PCs might be one of the reasons for the lack of LTP-IE in these cells.

Molecular mechanisms for the regulation of Kv1.2 have been extensively studied for the last two decades. Tyrosine phosphorylation of Kv1.2 elicits endocytosis of this channel (Nesti et al. 2004). The present study showed that PP2, a selective SFK inhibitor, blocks the intrinsic plasticity. We tried to examine possible involvement of PYK2 in the LTP-IE induction by testing LTP-IE in CA3-PCs transfected with shPYK2 (Hsin et al. 2010). Although the shPYK2-transfected CA3-PCs exhibited no sign of LTP-IE (Fig. 5E), it was noted that depletion of PYK2 itself makes the CA3-PCs more excitable (Fig. 5Eb, Supplemental Table 1). The activities of many ion channels and neurotransmitter receptors are regulated by tyrosine phosphorylation (reviewed in Ohnishi et al. 2011). Given that PYK2 is an upstream regulator of SFK, knock-down of PYK2 and the associated downregulation of SFK may have a profound influence on neuronal activity, which in turn may induce homeostatic plasticity (Cudmore et al. 2010). The accelerated AP onset time in shPYK2-transfected CA3-PCs implies that the baseline I_K(D) in these cells is already somewhat downregulated, which may be responsible for the lack of LTP-IE.

Comparison to previously known intrinsic plasticity

LTP-IE has been demonstrated in different brain regions using different induction protocols. The intrinsic properties that undergo activity-dependent plastic changes encompass a variety of passive and active electrical parameters depending on the biophysical properties of ion channels that mediate the LTP-IE. When a molecular mediator is high-voltage or Ca²⁺-activated channels, LTP-IE is associated with the changes in the AP shape or the firing rate potentiation without the change in input resistance (Nelson et al. 2003; Sourdet et al. 2003; Cudmore & Turrigiano, 2004; Nataraj et al. 2010). The LTP-IE shown in the present study, which is mediated by suppression of low-voltage-activated K_D channels, is different from previous studies, in that it is associated with an increase in the input resistance. Furthermore, to our knowledge, the induction protocol used in the present study is the mildest ever known except for the hyperpolarization-induced intrinsic plasticity in vestibular nucleus neurons (Nelson et al. 2003). We showed that LTP-IE could be induced by a brief train of somatic firing at the firing frequency recorded in vivo. Such a mild induction protocol contrasts with the strong non-synaptic induction protocols used in previous studies such as thousands of current-evoked APs (Cudmore & Turrigiano, 2004; Mahon & Charpier, 2012) or a long depolarization pulse for tens of seconds (Brown & Randall, 2009).

Cell signalling mechanisms that regulate the ion channel activity underlying intrinsic plasticity have been extensively studied for the last decade in CA1-PCs. The A-type K⁺ channel subunit (Kv4.2), SK and HCN channels have been implicated in the activity-dependent regulation of dendritic excitability in CA1-PCs (Frick et al. 2004; Fan et al. 2005; Kim et al. 2007; Hammond et al. 2008; Lin et al. 2008). Among these channels, HCN and Kv4.2 are expressed in the apical dendrite of CA1-PCs with higher density towards the distal dendrites (Hoffman et al. 1997; Lorincz et al. 2002). PKA and Ca²⁺/calmodulin-dependent kinase II (CaMKII) have been implicated in the trafficking of these channels, and thereby modulate dendritic excitability in CA1-PCs (Hammond et al. 2008; Lin et al. 2008). In addition, CaMKII involves the modulation of Na⁺ channel kinetics in CA1-PCs (Xu et al. 2005). We found no evidence for the involvement of these protein kinases in the LTP-IE induction in CA3-PCs (Fig. 5A and Supplemental Fig. 7), indicating that the LTP-IE induction in CA3-PCs does not involve intrinsic plasticity mechanisms that have been previously recognized in other types of neurons.

Dependence of LTP-IE induction on somatic firing frequency

The somatic activity-induced LTP-IE shown in the present study is distinct from the synaptic activity-induced intrinsic plasticity, in that the former is associated with a global increase in cellular excitability whereas the latter is associated with a local or branch-specific increase in dendritic excitability (Frick et al. 2004; Losonczy et al. 2008). Such changes in global excitability may facilitate the coupling of EPSP to spike (E–S coupling). Without being counter-balanced by complementary mechanisms, general potentiation of E–S coupling would lead to runaway dynamics of network activity subsequent to the enhanced propagation of neuronal activity and rapid saturation of information storage at Hebbian synapses. The complementary homeostatic regulation mechanisms might be necessary for counteracting the LTP-IE observed in the present study. It has been recognized that Kv1.2 undergoes reciprocal regulation by SFK and tyrosine phosphatase (Lev et al. 1995; Tsai et al. 1999; Nesti et al. 2004). We found that somatic firing at frequencies higher than 10 Hz induces rather smaller changes in G_in, implying that both of SFK and tyrosine phosphatase may be regulated by Ca²⁺ with different sensitivity. Involvement of tyrosine phosphatase in the enhancement of Kv1.2 and its upstream signalling molecules, however, remain to be elucidated in CA3-PCs.

Physiological implications

The LTP-IE shown in the present study requires intact distal apical dendrites and bAP-induced dendritic Ca²⁺ signalling (Fig. 7), indicating that the downregulation of Kv1.2 expressed in distal apical dendrites underlies the LTP-IE induction. This finding raises a possibility that repetitive somatic firing can alter not only spike timing (Cudmore et al. 2010; Higgs & Spain, 2011) but also the excitability of distal apical dendrites in CA3-PCs. In general, electrotonic attenuation of distal dendritic inputs to pyramidal neurons is compensated by high input resistance of the distal dendrites, activation of subthreshold Na⁺ current and dendritic spike generation (reviewed in Spruston, 2008). The latter two factors are opposed by dendritic K⁺ conductance. Therefore, K⁺ conductance in the distal dendrites may profoundly affect the E–S coupling of distal dendritic inputs and memory encoding at Hebbian synapses of the distal dendrites. Temporoammonic or perforant pathways, which convey important sensory information from the entorhinal cortex, innervate the distal apical dendrites of CA3-PCs, and make synapses that display Hebbian synaptic plasticity involving NMDA receptors (Debanne et al. 1998; McMahon & Barrionuevo, 2002). Such direct cortical inputs to CA3-PCs are thought to contribute to the retrieval of stored memory (Lee & Kesner, 2004). Therefore, the activity-dependent downregulation of Kv1.2 in distal apical dendrites may enhance the synaptic plasticity and E–S coupling of direct cortical inputs, which in turn may facilitate memory encoding and retrieval at distal dendritic synapses of CA3-PCs.

Glossary

4-AP

4-aminopyridine

AP

action potential

bAPs

back-propagating APs

CA1-PCs

CA1 pyramidal cells

CA3-PCs

CA3 pyramidal cells

CaTs

Ca²⁺ transients

DTX-I

dendrotoxin-I

DTX-K

dendrotoxin-K

G_in

input conductance

HCN channel

hyperpolarization-activated cation channel

I_K(D)

D-type K⁺ current

KO

knock-out

LTP-IE

long-term potentiation of intrinsic excitability

PTK

protein tyrosine kinase

PYK2

tyrosine phosphorylation involving proline-rich tyrosine kinase 2

RMP

resting membrane potential

SFKs

Src family kinases

SK channel

small-conductance Ca²⁺-activated K⁺ channel

VDCC

voltage-dependent Ca²⁺ channel

WT

wild-type

Additional information

Competing interests

The authors declare that they have no conflict of interest.

Author contributions

J.H.H. and S.-H.L. conceived and designed the experiments; J.H.H., K.E. and K.-H.L. performed the experiments; J.H.H., K.E., K.H.L. W.-K.H. and S.-H.L. analysed and interpreted the data; J.H.H., W.-K.H. and S.-H.L. wrote the paper. All authors have approved the final version of the manuscript.

Funding

This research was supported by a grant from the National Research Foundation of Korea (Grant No. 20120009135).

Supplementary material

Supporting Information