Abstract
Dendritic, backpropagating action potentials (bAPs) facilitate the induction of Hebbian long-term potentiation (LTP). Although bAPs in distal dendrites of hippocampal CA1 pyramidal neurons are attenuated when propagating from the soma, their amplitude can be increased greatly via downregulation of dendritic A-type K+ currents. The channels that underlie these currents thus may represent a key regulatory component of the signaling pathways that lead to synaptic plasticity. We directly tested this hypothesis by using Kv4.2 knock-out mice. Deletion of the Kv4.2 gene and a loss of Kv4.2 protein resulted in a specific and near-complete elimination of A-type K+ currents from the apical dendrites of CA1 pyramidal neurons. The absence of dendritic Kv4.2-encoded A-type K+ currents led to an increase of bAP amplitude and an increase of concurrent Ca2+ influx. Furthermore, CA1 pyramidal neurons lacking dendritic A-type K+ currents from Kv4.2 knock-out mice exhibited a lower threshold than those of wild-type littermates for LTP induction with the use of a theta burst pairing protocol. LTP triggered with the use of a saturating protocol, on the other hand, remained indistinguishable between Kv4.2 knock-out and wild-type neurons. Our results support the hypothesis that dendritic A-type K+ channels, composed of Kv4.2 subunits, regulate action potential backpropagation and the induction of specific forms of synaptic plasticity.
- K+ channels
- long-term potentiation
- hippocampus
- dendrites
- backpropagating action potential
- knock-out mouse
Introduction
Correlated presynaptic and postsynaptic activity results in long-term changes in synaptic strength. The phenomenon is known as Hebbian-type plasticity (Magee and Johnston, 1997; Markram et al., 1997; Bi and Poo, 1998; Debanne et al., 1998; Sjostrom et al., 2001; Golding et al., 2002; Watanabe et al., 2002). A common factor in the induction of Hebbian-type long-term potentiation (LTP) is the necessity of postsynaptic depolarization, which facilitates NMDA receptor activation and the activation of voltage-dependent Ca2+ channels, both of which lead to a Ca2+ rise in the postsynaptic neuron (Magee and Johnston, 1997; Golding et al., 2002; Frick et al., 2004; Lisman and Spruston, 2005). One of the physiological mechanisms of postsynaptic depolarization is the backpropagating action potential (bAP), which usually is generated in the axon and backpropagates into the dendrites (Stuart and Sakmann, 1994; Spruston et al., 1995). Pairing of bAPs and EPSPs can result in supralinear summation in the dendrites of membrane voltage (Magee and Johnston, 1997; Stuart and Hausser, 2001) and of Ca2+ influx (Yuste and Denk, 1995; Magee and Johnston, 1997; Koester and Sakmann, 1998; Mainen et al., 1999; Nevian and Sakmann, 2004), leading to the induction of LTP (Magee and Johnston, 1997; Markram et al., 1997; Watanabe et al., 2002; Frick et al., 2004; Fan et al., 2005). bAPs in the dendrites of hippocampal CA1 pyramidal neurons are an essential feature for the induction of LTP of the Schaffer collateral pathway under physiological conditions (Magee and Johnston, 1997). It is therefore conceivable that regulators of bAPs also regulate the induction of LTP.
In the CA1 pyramidal neurons of hippocampus the amplitude of these bAPs decreases as they travel farther away from the soma (Spruston et al., 1995; Magee and Johnston, 1997; Yuan et al., 2002; Bernard and Johnston, 2003; Frick et al., 2004), despite the uniform density of voltage-dependent Na+ channels (Magee and Johnston, 1995). In contrast, the density of the A-type K+ currents increases along the apical dendrites (Hoffman et al., 1997). The location and properties of the A-type K+ currents make them suitable as key regulators of action potential backpropagation (Hoffman and Johnston, 1998; Watanabe et al., 2002; Yuan et al., 2002; Chen and Johnston, 2004).
A-type K+ currents in CA1 pyramidal neurons can be encoded by either the Kv1.4 or Kv4.2 gene (Coetzee et al., 1999). However, based on immunohistochemistry studies, the Kv1.4 proteins are found primarily in the axon, whereas the Kv4.2 proteins are found mostly in the dendrites (Sheng et al., 1992; Gu et al., 2003). Consistent with this localization and in the context of synaptic plasticity in CA1 pyramidal neurons, the threshold of LTP induction of field EPSPs is lowered by heteropodatoxin 3 (HpTX3), a relatively selective blocker of Kv4 channels (Ramakers and Storm, 2002).
In this report, we investigated a key test of the overall hypothesis linking dendritic A-type K+ current, action potential backpropagation, and LTP induction by evaluating the physiology of CA1 pyramidal neurons in mice specifically lacking the Kv4.2 gene. Previous studies show that transient K+ currents in cardiac myocytes and spinal cord dorsal horn neurons are reduced in these Kv4.2−/− mice (Jung, 2002; Guo et al., 2005; Hu et al., 2006). This report focuses on the CA1 region of hippocampus.
Materials and Methods
Transgenic animals and genotyping
Kv4.2−/− mice were generated in 129/SvEv background (Guo et al., 2005). Littermate genotypes were confirmed by PCR results on the basis of Kv4.2-specific primers (forward, GTG GAT GCC TGT TGC TTC; reverse, CCC ACA AGG CAG TTC TTT TA) and neo-specific primers (forward, AGG ATC TCC TGT CAT CTC ACC TTG CTC CTG; reverse, AAG AAC TCG TCA AGA AGG CGA TAG AAG GCG).
Preparation of acute hippocampal slices from mouse brain
Hippocampal slices were prepared from 5- to 12-week-old mice by following standard procedures. Briefly, the animals were anesthetized by lethal dose of a mix of ketamine and xylazine and perfused through the heart with ice-cold cutting solution containing the following (in mm): 240 sucrose, 2.5 KCl, 1.25 NaH2PO4, 25 NaHCO3, 0.5 CaCl2, and 7 MgCl2, saturated with 95% O2/5% CO2 before freezing. Both hemispheres then were removed quickly and sliced at 350 μm in thickness with a vibratome. After incubation in a holding chamber for at least 30 min at room temperature, the slices were transferred into the recording chamber. Both the holding and the recording chambers contained the following (in mm): 125 NaCl, 2.5 KCl, 1.25 NaH2PO4, 25 NaHCO3, 2.0 CaCl2, 1.0 MgCl2, and 25 dextrose.
Organotypic culture of mouse hippocampus
Mouse organotypic slices were prepared and cultured according to the interface technique (Stoppini et al., 1991). Slices were prepared from postnatal day 1–7 mice at 325 μm thickness with a McIlwain tissue chopper (Mickle Laboratory Engineering, Guildford, Surrey, UK) and cultured on Millicell inserts (Millipore, Bedford, MA) in a MEM-based medium (Invitrogen, Carlsbad, CA) containing 20% horse serum (Invitrogen). Medium was changed every 2 d, and the slices could be kept alive for at least 4 weeks in culture.
Electrophysiological recordings
A Zeiss Axioskop (Oberkochen, Germany) fit with a 40 or 60× water immersion objective and differential interference contrast (DIC) was used to view the slices. Light in the near-infrared (IR) range (740 nm) in conjunction with a contrast-enhancing camera was used to visualize individual dendrites. An Axopatch-200 or Axopatch-1D amplifier was used for voltage-clamp recordings. Current-clamp recordings were performed by using either an Axoclamp-2A or a BVC700A amplifier. Whole-cell recording pipettes (7–12 MΩ) contained the following (in mm): 120 KMeSO4 or 120 K-gluconate, 20 KCl, 10 HEPES, 0.2 EGTA, 8 NaCl, 4 Mg-ATP, 0.3 Tris-GTP, and 14 phosphocreatine (pH 7.25-adjusted with KOH). For cell-attached patch recordings the pipettes (10–14 MΩ) were wrapped with Parafilm to reduce pipette capacitance; the tips were inspected visually for uniform diameter (∼1 μm) and were fire polished. Pipette solution for cell-attached recording of voltage-dependent K+ channels contained the following (in mm): 125 NaCl, 10 HEPES, 2 CaCl2, 1 MgCl2, and 2.5 KCl plus 1 μm TTX (pH 7.4-adjusted with KOH). To isolate whole-cell voltage-dependent K+ currents in pyramidal neurons, we added 1 μm TTX and 2 mm MnCl2 to the bath solution with 0 Ca2+ and 0 NaH2PO4 to block Na+ currents, Ca2+ currents, and Ca2+-activated K+ currents. Pulse generation and data acquisition were controlled with custom software written in the IgorPro environment. Linear leak and capacitive currents were subtracted digitally by using null traces or scaled traces of smaller voltage steps.
To induce LTP, we paired subthreshold synaptic stimulation and postsynaptic action potentials (APs). Schaffer collateral fibers were stimulated electrically with a bipolar microelectrode placed in stratum radiatum at a lateral distance of 50–100 μm from the apical dendrite. Axonally initiated APs were elicited with somatic depolarizing current injection or antidromic stimulation via a stimulating electrode in the alveus. Test stimuli were delivered every 20–30 s; a hyperpolarizing current pulse was injected into the cell after the test stimulus to monitor input resistance and series resistance. Slope measurements of EPSP were made from a line fit to the rising phase of the EPSP. The magnitude of potentiation was quantified by averaging the slope of EPSP rising phase for a 10 min period between 20 and 30 min after induction and dividing this number by the average slope of EPSP rising phase during the 10 min baseline before induction. During LTP induction the number of APs actually fired by the cell did not always match the number of stimuli (current injections or antidromic stimuli). The actual number of APs fired by the cell was counted, and an average per burst was calculated for every experiment.
Whole-cell voltage-clamp recordings in organotypic slice culture were performed at room temperature (22–24°C). In acute slices the cell-attached voltage-clamp recordings were performed at room temperature (22–24°C); whole-cell current-clamp recordings were performed at near-physiological temperature (32–34°C). Dendritic recordings were made from the trunk of the apical dendrites. We performed whole-cell voltage-clamp experiments in organotypic slice cultures because cells in this preparation are electrotonically more compact than cells in acute slices. This allows us to have a better estimation of the amplitude of whole-cell current with voltage clamp (Johnston and Brown, 1983; Jeromin et al., 2003). We performed all of the voltage-clamp experiments at room temperature instead of near-physiological temperature because of better stability of the recordings (in both the whole-cell and the cell-attached configuration). With the whole-cell configuration this also allows for better voltage clamp because of slower kinetics (Johnston and Brown, 1983).
Ca2+ imaging
We performed Ca2+ fluorescence imaging in a similar way to that described previously (Frick et al., 2003). A Quantix 57 CCD camera (Roper Scientific, Duluth, GA) with a 535 × 512 pixel array and single wavelength (380 nm) excitation was used, with changes in [Ca2+]i quantified by calculating ΔF/F, where F is the fluorescence intensity before stimulation (after subtracting autofluorescence) and ΔF is the change in fluorescence during neuronal activity (corrected for bleaching). The autofluorescence of the tissue was measured in a region of equal size but adjacent to the dye-filled neuron, either in the dendritic field or in the cell body layer, and bleaching was determined by measuring the change in fluorescence at rest (without stimulation). The ΔF/F measurements usually were repeated four to six times and averaged. Sequential frame rate was 50–100 Hz, and pixels were binned in a 5 × 5 array. To reach the equilibrium state of dye diffusion, we typically waited at least 20 min after break-in to allow the indicator dye (bis-fura-2) to diffuse into the most distal part of the cell before we recorded optical signals. The distance of any given location along an apical dendrite was measured from the base of apical dendrites. Ca2+ imaging experiments were performed at near-physiological temperature (32–34°C).
Wild-type controls
For organotypic slice culture the non-littermate wild-type 129/SvEv mice were used as controls. Gender of the animals could not be differentiated at the time of culture.
For experiments in acute slices both non-littermate and littermate wild-type 129/SvEv mice were used as controls in recordings of dendritic K+ currents, bAPs, and Ca2+ signals. Age-matched littermate controls were used for LTP experiments. Male mice were used for experiments in acute slices.
Biochemical analysis
Sample preparation.
Adult mice (129S6/SvEvTaq or Kv4.2 knock-out mice) were decapitated, and their brains were placed rapidly into ice-cold cutting saline [containing the following (in mm): 110 sucrose, 60 NaCl, 3 KCl, 1.25 NaH2PO4, 28 NaHCO3, 7 MgCl2, 0.5 CaCl2, 5 glucose, and 0.6 ascorbate, saturated with 95% O2/5% CO2]. Then the brain regions of interest were removed, immediately frozen on dry ice, and stored at –80°C until assayed.
The hippocampi were homogenized and briefly sonicated in ∼600 μl of ice-cold homogenization buffer [containing the following (in mm): 20 Tris, pH 7.5, 1 EGTA, 1 EDTA, 1 Na4P2O7, 4 para-nitrophenylphosphate, and 1 sodium orthovanadate plus 100 μm phenylmethylsulfonyl fluoride and 10 μl/ml protease inhibitor mixture (Sigma-Aldrich, St. Louis, MO)]. Membrane fractions were prepared. The samples were centrifuged at 1000 × g for 5 min to remove cellular debris; then the supernatant was centrifuged at 100,000 × g for 30 min. The pellet was resuspended in 5% SDS in homogenization buffer. The protein content of each sample was assessed (DC protein assay; Bio-Rad, Hercules, CA), and SDS/homogenization buffer was added to normalize the protein level between samples. Sample buffer (0.3 m Tris, pH 6.8, 40% glycerol, 8% SDS, and 150 mm dithiothreitol) was added, and the samples were stored at –20°C. Samples were loaded onto 10% SDS-polyacrylamide gels and resolved by standard electrophoresis. Gels were transferred electrophoretically to Immobilon polyvinylidene difluoride membranes.
Western blotting.
Immobilon filters were blocked for 1 h at room temperature in 5% dry milk and 3% bovine serum albumin dissolved in Tween–Tris-buffered saline (TTBS; 50 mm Tris, pH 7.5, 150 mm NaCl, 1 μm microcystin, 0.1% Tween 20) or 0.2% I-Block in TTBS. The blots were incubated for either 1 h at room temperature or overnight at 4°C with the primary antibody and then were washed four times (10 min each) with TTBS. Blots were incubated with an HRP-conjugated anti-rabbit (1:20,000; Cell Signaling Technology, Beverly, MA) or anti-mouse (1:20,000; Cell Signaling Technology) secondary antibody and then washed four times with TTBS. The blots were developed with Enhanced Chemiluminescence (Amersham Biosciences, Arlington Heights, IL) and visualized on film [Kodak BioMax MR (Rochester, NY) or ISC BioExpress Blue Lite (Kaysville, UT)]. Primary antibodies that were used included Kv4.3, Kv1.4, Kv3.4, and the H channel proteins HCN1 and HCN2 (1:500, 1:500, 1:200, 1:500, 1:500, respectively; Chemicon, Temecula, CA); glutamate receptors GluR1 and GluR2/3 (1:1000; Upstate Biotechnology, Lake Placid, NY); K+ channel-interacting proteins KChIP1, KChIP2, and KChIP3 (1:3500, 1:2500, 1:100, respectively; gifts from Dr. J. Trimmer, University of California, Davis, CA); postsynaptic density-95 (PSD-95; 1:1000; Zymed, San Francisco, CA); synapse-associated protein 97 (SAP97; 1:3000; Affinity BioReagents, Neshanic Station, NJ); neuronal calcium sensor protein 1 (NCS1; 1:1000; Santa Cruz Biotechnology, Santa Cruz, CA); dipeptidyl peptidase X (DPPX; 1:500; a gift from Dr. W. Cookson, Oxford University, Oxford, UK); and pentraxin 3 (PTX3; 1:1000; Axxora, San Diego, CA). Blots were stripped of the original antibody by being washed twice with 0.2 m NaOH (20 min each) and then four times in TTBS. Western blotting was repeated on the same blots with an anti-actin antibody (1:1000; Sigma–Aldrich) to ensure equal loading between lanes. Densitometric analysis of immunoreactivity was conducted with a desktop scanner and NIH ImageJ software (Bethesda, MD).
Statistical analysis
Significance (p < 0.05) was determined with two-sample Student's t tests. Error bars represent the SEM.
Results
Deleting the Kv4.2 gene led to abolishment of Kv4.2 protein in the brain and spinal cord
The Kv4.2 knock-out (Kv4.2−/−) mouse was generated by targeted disruption of exon I of the Kv4.2 gene (Guo et al., 2005). We performed Western blot analysis on total protein extracted from the cortex, hippocampus, cerebellum, and spinal cord of Kv4.2−/−. Tissue from age-matched wild-type mice (Kv4.2+/+) of the same genetic background was used as control. As expected, the Kv4.2 protein was detected at high levels in various brain regions of control mice but was undetectable in Kv4.2−/− mice (Fig. 1A).
Absence of Kv4.2 protein and reduced expression level of KChIPs in Kv4.2−/− mice. A, Western blotting confirmed the absence of Kv4.2 protein in various CNS regions of Kv4.2−/− mice, including cortex, hippocampus, cerebellum, and spinal cord [n = 9 of wild-type (WT); n = 9 of Kv4.2−/− (KO)]. B, The absence of Kv4.2 protein was accompanied by a significant decrease of KChIPs in Kv4.2−/− mice. Kv4.3: n = 9 of wild-type, n = 8 of Kv4.2−/−; Kv1.4, HCN1, and HCN2: n = 9 of wild-type, n = 9 of Kv4.2−/−; Kv3.4: n = 7 of wild-type, n = 7 of Kv4.2−/−; GluR1 and GluR2/3: n = 7 of wild-type, n = 7 of Kv4.2−/−; SAP97, PSD-95, and NCS1: n = 4 of wild-type, n = 4 of Kv4.2−/−; DPPX and PTX3: n = 8 of wild-type, n = 8 of Kv4.2−/−; KChIP1, KChIP2, and KChIP3: n = 7 of wild-type, n = 7 of Kv4.2−/−. Error bars represent the SEM; *p < 0.05.
Expression levels of 15 other proteins in the hippocampus also were analyzed with Western blotting, including other K+ channel proteins potentially underlying neuronal A-type K+ currents: Kv4.3, Kv1.4, and Kv3.4; the H channel proteins HCN1 and HCN2; the AMPA-type glutamate receptor proteins GluR1 and GluR2/3; two membrane-anchoring proteins enriched in the postsynaptic area, SAP97 and PSD-95; and proteins that functionally or physically associate with Kv4.2 such as NCS1, DPPX, a secretory-type protein containing a pentraxin domain (PTX3), and KChIP1, KChIP2, and KChIP3 (Rhodes et al., 2004; Menegola and Trimmer, 2006). The protein levels of KChIP1, KChIP2, and KChIP3 were reduced significantly, by 30–70%, in the hippocampus of Kv4.2−/− mice as compared with wild type. Levels of the other proteins were not altered significantly in Kv4.2−/− (Fig. 1B). These data confirm a complete loss of CNS Kv4.2 protein in Kv4.2 knock-out animals. Moreover, these findings suggest an interesting coordinate regulation of Kv4.2 and KChIP1, KChIP2, and KChIP3 (Menegola and Trimmer, 2006).
Differential changes in the Kv1- and Kv4-mediated components of whole-cell A-type K+ currents in Kv4.2−/−
Both the Kv1 and Kv4 families of K+ channel genes are expressed in CA1 pyramidal neurons. Kv1- and Kv4-mediated A-type K+ currents differ in the kinetics of recovery from inactivation (Coetzee et al., 1999). The time course of recovery from inactivation of Kv4 channels and dendritic A-type current (τ < 100 ms) (Johnston et al., 2000; Jerng et al., 2004) is faster than that of Kv1 channels (τ > 1 s) (Castellino et al., 1995) by an order of magnitude. Hence the relative contributions of Kv1 channels and Kv4 channels to whole-cell A-type K+ currents can be studied with a voltage protocol that takes advantage of their difference in recovery from inactivation.
Figure 2 illustrates such a protocol. Whole-cell voltage-clamp recordings were performed from the soma of CA1 pyramidal neurons in organotypic slice cultures. The membrane potential was held at −80 mV, and two identical voltage steps to +20 mV with a duration of 200 ms and interval of 100 ms were applied to activate voltage-dependent K+ channels. In wild-type neurons the first depolarizing step activated whole-cell K+ currents with an A-type component that fully inactivated at the end of the 200 ms step. During the 100 ms hyperpolarization between the two depolarizing steps a large portion of the A-type current recovered from inactivation. The amplitude of the first transient (peak current subtracted by the non-inactivating component) represents the total amount of whole-cell A-type current. The amplitude of the second transient represents the amount of whole-cell A-type current that is mediated by Kv4 channels. The difference between the first transient and the second transient represents the amount of Kv1-mediated current that would not be expected to recover from inactivation during the 100 ms hyperpolarization (Varga et al., 2004).
Reduction of Kv4-mediated and upregulation of Kv1-mediated whole-cell A-type K+ current in Kv4.2−/−. Whole-cell voltage-clamp recordings were performed from the soma of CA1 pyramidal neurons in organotypic slice cultures. A, CA1 pyramidal neurons from wild-type (WT) mice exhibited a prominent portion of the transient A-type K+ current, which recovered from inactivation to 82.2% within 100 ms, indicating that >80% of native A-type current was from Kv4. B, Under the same condition there was a significant reduction of the first transient amplitude and an even further reduction of the second transient amplitude in Kv4.2−/− (C). D, As a result, there was a <20% recovery of A-type current within 100 ms in Kv4.2−/−. The difference between the first and the second transient amplitude reflected the amount of transient current that did not recover from inactivation within 100 ms and, thus, was not mediated by Kv4 subunits. E, The non-Kv4-mediated current was upregulated significantly in Kv4.2−/−. The sustained component of K+ currents in CA1 pyramidal neurons remained the same between WT and Kv4.2−/− mice. Error bars represent the SEM; **p < 0.01; ***p < 0.005.
In comparison to the wild-type neurons (39.14 ± 3.30 pA/pF; n = 6), CA1 pyramidal neurons of Kv4.2−/− had a significantly reduced amount of total A-type K+ currents (21.97 ± 1.54 pA/pF; n = 10). The Kv4-mediated A-type current (as revealed by the second voltage step), in particular, was mostly absent in Kv4.2−/− neurons (4.94 ± 0.68 pA/pF in knock-out as compared with 32.06 ± 2.41 pA/pF in wild type). However, the A-type current component that did not recover within 100 ms, quantified as the difference between the first and the second transient, was 7.08 ± 1.02 pA/pF (n = 6) in wild-type and 17.03 ± 1.74 pA/pF (n = 10) in CA1 pyramidal neurons of Kv4.2−/−. The whole-cell capacitance was not different between wild-type (26.5 ± 2.3 pF; n = 6) and Kv4.2−/− (29.2 ± 2.1 pF; n = 10) neurons. These results suggest that, although total Kv1.4 protein amount appears to be unchanged in whole hippocampus in Kv4.2−/− (Fig. 1), some non-Kv4 subunits, most likely Kv1 family members, are upregulated in Kv4.2−/−, at least in CA1 neurons (Fig. 2C–E).
Dendritic A-type K+ current was eliminated specifically and completely in Kv4.2−/−
We next asked whether the dendritic A-type K+ current was affected by the loss of Kv4.2. Cell-attached recordings of macroscopic voltage-dependent K+ currents were made from the dendrites of Kv4.2−/− and wild-type mice in acute slices. The results are illustrated in Figure 3. In CA1 dendrites the macroscopic K+ currents usually are composed of a transient A-type component and a sustained component (Hoffman et al., 1997), and the amplitude of both components varies considerably (Chen and Johnston, 2004). In the Kv4.2−/− dendrites, however, macroscopic K+ currents did not have a transient component that could be measured reliably (Fig. 3A). In wild-type dendrites 200–300 μm from soma the peak A-type current had an amplitude of 15.0 ± 4.1 pA (distance from soma, 225 ± 9 μm; n = 15), whereas in Kv4.2−/− dendrites at comparable distances from the soma the peak transient had an amplitude of 1.4 ± 0.5 pA (distance from soma, 241 ± 8 μm; n = 11; p = 0.005) (Fig. 3B). The sustained component of macroscopic K+ currents did not seem to be different between wild-type (6.7 ± 1.4 pA; n = 15) and Kv4.2−/− dendrites (5.2 ± 1.0 pA; n = 11; p = 0.381) (Fig. 3C). Thus A-type K+ current was eliminated specifically in dendrites of CA1 pyramidal neurons of Kv4.2−/−. Although compensation from the upregulation of other Kv channel-mediated A-type current was detected at the whole-cell level (Fig. 2), such compensation did not appear to occur in the apical dendrite.
Cell-attached recordings revealed the absence of dendritic A-type K+ current in CA1 pyramidal neurons of Kv4.2−/−. Cell-attached voltage-clamp recordings were performed from the trunk of the apical dendrites of CA1 pyramidal neurons in acute slices. A, In wild-type (WT) mice there was prominent A-type current in the apical dendrites (200–300 μm from soma) of CA1 pyramidal neurons, which we recorded as the transient component of total K+ current with a voltage step from −100 to +50 mV. This component was missing in the dendrites of Kv4.2−/− mice. B, In WT dendrites, the amplitude of A-type current was 15 ± 4.1 pA (n = 15; 11 from non-littermate controls and 4 from littermate controls). In Kv4.2−/− the amplitude of A-type current was 1.4 ± 0.5 pA (n = 11; 7 from non-littermate Kv4.2−/− and 4 from littermate Kv4.2−/−). C, The amplitude of the sustained K+ current was not altered in dendrites of Kv4.2−/−. In WT dendrites the amplitude of the sustained current was 6.7 ± 1.4 pA (same cells as in B). In Kv4.2−/− the amplitude of the sustained current was 5.2 ± 1.0 pA (same cells as in B).
bAPs in dendrites were of larger amplitude in Kv4.2−/−
To test whether the lack of dendritic A-type current affected dendritic APs, we recorded bAPs in the dendrites of Kv4.2−/− CA1 neurons and wild-type controls. Figure 4A shows examples of recordings in the soma and in dendrites 250 μm from the soma. Recordings were performed at near-physiological temperature (32–34°C). When recorded in the range of 220–300 μm along the dendrite in wild-type neurons (distance from soma, 238 ± 4 μm; n = 19), the amplitude of the bAPs was 29.8 ± 1.2 mV (n = 19). At comparable locations (distance from soma, 246 ± 6 μm; n = 15) and temperature the bAP amplitude (57.9 ± 2.6 mV; n = 15) was significantly larger in Kv4.2−/− than in wild-type dendrites (p ≪ 0.001) (Fig. 4A,B).
Elimination of dendritic A-type K+ current resulted in larger dendritic bAPs in Kv4.2−/− mice. Whole-cell current-clamp recordings were performed from CA1 pyramidal neurons in acute slices. A, Comparison of bAP amplitude between wild-type (WT) and Kv4.2−/−. APs recorded from soma showed no significant difference between Kv4.2−/− and WT controls. At 250 μm from soma the amplitude of bAP was larger in Kv4.2−/− as compared with WT. B, At comparable distances from the soma (220–300 μm) the amplitude of bAP was 29.8 ± 1.2 mV in WT dendrites (n = 19; 15 from non-littermate controls and 4 from littermate controls) and 57.9 ± 2.6 mV in Kv4.2−/− (n = 15; 11 from non-littermate Kv4.2−/− and 4 from littermate Kv4.2−/−). ***p < 0.005.
Additional analysis of bAPs revealed no difference in initial dV/dt between Kv4.2−/− (21.1 ± 2.8 mV/ms; n = 14) and wild-type neurons (20.6 ± 1.8 mV/ms; n = 16; p = 0.86), suggesting similar Na+ channel activation (Colbert and Johnston, 1996). Maximum dV/dt was slightly larger in Kv4.2−/− [31.2 ± 2.6 mV/ms (n = 14) as compared with 26.1 ± 1.9 mV/ms in wild-type (n = 16)], but the difference was not statistically significant (p = 0.12). Half-width of bAPs was slightly shorter in Kv4.2−/− [4.9 ± 0.4 ms (n = 14) as compared with 6.0 ± 0.5 ms in wild-type (n = 16); p = 0.11], possibly as a consequence of the larger amplitude of bAPs that reached the activation threshold for other types of dendritic K+ channels to participate in repolarization (Bernard and Johnston, 2003; Chen and Johnston, 2006).
Concurrent Ca2+ influx during bAPs was larger in Kv4.2−/−
bAPs lead to Ca2+ influx in the dendrites via voltage-dependent Ca2+ channels (Yuste and Denk, 1995; Magee and Johnston, 1997). With larger bAPs propagating in the apical dendrites of CA1 pyramidal neuron of Kv4.2−/− mice, we tested whether this increase in voltage change in dendrites is translated into a similar increase of Ca2+ influx. CA1 pyramidal neurons were filled with bis-fura-2 (100 μm) during somatic whole-cell recordings. The amplitude of Ca2+ signals at multiple dendritic locations in the same neuron was monitored and quantified as ΔF/F while single APs were initiated with somatic current injection. Because the surface–volume ratio varies along the dendrites, the ΔF/F value as a function of dendritic location usually peaks at the proximal part of the apical dendrites (Regehr and Tank, 1992; Jaffe et al., 1994; Schiller et al., 1995). As previously described (Frick et al., 2003) in rat hippocampus, bAPs resulted in concurrent Ca2+ influx. The amount of Ca2+ rise at more distal dendritic locations (beyond 125 μm) decreased with distance from soma, in parallel with the decrement of bAP amplitude (wild type; n = 8) (Fig. 5). We chose 125 μm as the location to which we normalized the ΔF/F of various locations. Because bAPs were larger in Kv4.2−/− than in wild-type dendrites, Ca2+ signals should have less decrement along the dendrites of Kv4.2−/− neurons. This is indeed what we observed with Ca2+ imaging (Fig. 5). In Kv4.2−/− the bigger bAP amplitudes appeared to result in more Ca2+ influx (knock-out; n = 7) (Fig. 5).
bAP-evoked Ca2+ signal in dendrites declined less with distance in Kv4.2−/−. Whole-cell current-clamp recordings were performed from CA1 pyramidal neurons in acute slices. A, Apical dendrites of CA1 pyramidal neurons were filled with 100 μm bis-fura-2. B, In response to a brief current injection (2 nA for 2.5 ms) the neuron fired a single AP. Relative change of intracellular Ca2+ was quantified as ΔF/F at various locations on dendrites after bleach correction and autofluorescence subtraction. In wild-type (WT) ΔF/F declined at ∼150 μm and almost reached 0 in distal dendrites (∼250 μm). In contrast, the value in Kv4.2−/− stayed relatively high. C, Normalized ΔF/F to that of dendritic location 125 μm was plotted as a function of distance from soma. Ca2+ influx caused by a single bAP in Kv4.2−/− (n = 7; 6 from non-littermate Kv4.2−/− and 1 from littermate Kv4.2−/−) was much larger than that of WT (n = 8; 3 from non-littermate controls and 5 from littermate controls), and this difference was more obvious in distal dendrites. Error bars represent SEM.
Induction threshold of Hebbian-type LTP was lowered in Kv4.2−/−
bAPs and concurrent Ca2+ influx in the dendrites are required for the induction of LTP with the theta burst pairing (TBP) protocol at the Schaffer collateral synapses of CA1 neurons (Magee and Johnston, 1997). The TBP protocol involves pairing of theta burst stimulation with postsynaptic current injection or antidromic stimulation to induce APs (Fig. 6A). The theta burst stimulation consists of five stimuli of the Schaffer collaterals at 100 Hz per burst and 10 repeats of the burst at the theta frequency (5 Hz) (Fig. 6B). A standard protocol to induce LTP reliably has each of the five EPSPs in every burst paired with a postsynaptic AP. In each pair of an EPSP plus AP, the stimulus for the EPSP usually precedes the stimulus for the AP by 7–8 ms (Frick et al., 2004; Fan et al., 2005). We used this protocol to try to induce LTP in CA1 neurons from Kv4.2−/− and their wild-type littermates. As is shown in Figure 6C, pairing the five EPSPs in every burst with three to five bAPs induced a similar magnitude of LTP in Kv4.2−/− (2.63 ± 0.35; n = 7) and wild-type littermate controls (2.53 ± 0.28; n = 9). This is consistent with previous studies that found no significant difference between Kv4.2−/− and control animals with respect to the amount of maximum LTP (Jung, 2002). Field EPSPs were recorded at room temperature in those studies, and a variety of LTP induction protocols were used (Jung, 2002).
LTP induction was different in Kv4.2−/− and control littermates. Whole-cell current-clamp recordings were performed from CA1 pyramidal neurons in acute slices. A, Experimental configuration. EPSPs were evoked by extracellular stimulation of Schaffer collateral pathways (S.C.). Postsynaptic APs were elicited either by brief, intracellular current injection or antidromic stimulation of pyramidal neuron axons. B, TBP protocol in which one AP was paired with the fifth EPSP during a burst of five EPSPs. The five EPSPs within the burst were at 100 Hz. A train of stimuli consisted of 10 repeats of such bursts at 5 Hz. Five such trains of stimulation were applied repeatedly at a 12 s interval. The number of bAPs that paired with the EPSPs was varied in the experiments. C, In response to a strong pairing protocol in which three to five APs were paired with five EPSPs in every burst, persistent LTP was induced in wild-type (WT; n = 9) and Kv4.2−/− (n = 7) mice. D, In response to a relatively weak pairing protocol in which fewer than two APs were paired with EPSPs, LTP was of significantly larger magnitude in Kv4.2−/− (n = 9) than in control littermates (n = 19).
It is known that, with TBP, LTP occurs only when sufficient numbers of EPSPs and bAPs are paired in time (Watanabe et al., 2002). We reasoned that, because bAPs were larger in Kv4.2−/−, these bAPs should be more effective than those of wild-type neurons in unblocking NMDA receptors and inducing LTP. As a consequence, fewer postsynaptic APs would be needed for LTP induction. We therefore adjusted the strength of induction in the TBP protocol by varying the number of APs that were paired with EPSPs.
When fewer than two APs on average were paired with EPSPs in a burst, LTP induction was significantly more robust in Kv4.2−/− (2.31 ± 0.38; n = 9) than in their wild-type littermates (1.27 ± 0.13; n = 19) (Figs. 6D, 7A). We consider >50% increase in the initial EPSP slope after 20 min of TBP as successful LTP. By this criterion, the likelihood of LTP induction with this smaller number of pairings was >60% in Kv4.2−/− and ∼20% in wild-type (Fig. 7A). We further grouped the data according to the average numbers of APs being paired with EPSPs in a burst and examined the amount of LTP in each group (Fig. 7B). LTP could be induced in Kv4.2−/− when less than one AP on average was paired with the EPSPs (1.72 ± 0.29; n = 4). In wild-type littermate controls when the number of paired APs was minimal (0.5–0.9 per burst on average), EPSP initial slope remained mostly unchanged after induction (1.00 ± 0.13; n = 5). Thus in wild-type animals LTP could be induced only when >1.5 APs on average were paired with EPSPs. Specifically, 1 to 1.9 AP and 2 to 2.9 AP induction protocols resulted in 1.36 ± 0.16-fold (n = 14) and 1.94 ± 0.33-fold (n = 5) changes in EPSP initial slope in wild-type neurons, respectively. LTP magnitude increased as the number of paired APs increased and reached its maximal level at three to five APs (2.53 ± 0.28; n = 9). However, the amount of LTP appeared already to exhibit a maximum plateau level with two APs in Kv4.2−/− (2.78 ± 0.6-fold change in EPSP slope; n = 5). If we operationally define the “induction threshold of LTP” as the required minimum numbers of APs that are paired with EPSPs, then in Kv4.2−/− the induction threshold of LTP was significantly lower than in wild-type animals.
The induction threshold of LTP was lower in Kv4.2−/− than in control littermates. A, Cumulative probability plot showing higher probability of LTP induction in Kv4.2−/− than in control littermates, when a weak (<2 AP–EPSP pairings) TBP protocol was used. The dotted line indicates a 50% increase of EPSP slope. B, Amount of LTP was plotted against the number of AP–EPSP pairings and compared between Kv4.2−/− and wild-type (WT) littermate controls. Pairing fewer than two APs (0.5–0.9 and 1–1.9 groups) with EPSPs was sufficient to induce LTP in Kv4.2−/−, but not in control, animals. Stronger protocols consisting of two to 2.9 APs or three to five APs resulted in robust LTP in both genotypes, and the amount of LTP was not statistically different. Error bars represent the SEM; *p < 0.05; ***p < 0.005; n.s., not significant, p > 0.05. The number of cells in each group is indicated in the bars of the graph.
Discussion
Many types of cells in a variety of species express a voltage-dependent K+ current with rapid activation and inactivation (Coetzee et al., 1999). These transient K+ currents collectively are called the “A-type K+ currents.” Their importance in electrical signaling has motivated significant research effort in elucidating their molecular identities in different types of cells (Guo et al., 2005; Hu et al., 2006).
Several lines of evidence point to the A-type K+ current as a key regulator of action potential backpropagation in the CA1 pyramidal neurons. First, the progressive decrease of bAP amplitude parallels the increase of A-type K+ current density along the length of the apical dendrites (Hoffman et al., 1997). Second, the amplitude of dendritic bAP is increased greatly by 5–10 mm 4-aminopyridine (Hoffman et al., 1997). Although a nonspecific drug, 4-aminopyridine preferentially blocks the A-type K+ current in CA1 neurons (Storm, 1990; Hoffman et al., 1997). Third, activation of cAMP-dependent protein kinase (PKA) or protein kinase C (PKC) downregulates A-type K+ current by shifting the activation curve to a more depolarized voltage range (Hoffman and Johnston, 1998). Activation of PKA or PKC also increases the amplitude of dendritic bAPs (Hoffman and Johnston, 1998; Yuan et al., 2002). Similarly, inhibition of the MAP (mitogen-activated protein) kinase pathway upregulates A-type K+ current by shifting the activation curve to a more hyperpolarized voltage range (Watanabe et al., 2002). The same manipulation results in a decrease of bAP amplitude (Watanabe et al., 2002; Yuan et al., 2002) and suppression of LTP induction (Watanabe et al., 2002). The dendritic A-type K+ current also underlies changes in local dendritic excitability during the expression of LTP (Frick et al., 2004). In addition, the dendritic A-type K+ current appears to be downregulated in an animal model of temporal lobe epilepsy, a condition in which bAPs are found to have less decrement in their dendrites (Bernard et al., 2004).
The Kv4.2 gene encodes a transient A-type channel in heterologous expression systems (Baldwin et al., 1991), and, immunohistochemically, Kv4.2 proteins are located preferentially in the dendrites of hippocampal neurons (Sheng et al., 1992). Hence we hypothesized that the Kv4.2 gene encodes the pore-forming α-subunit of dendritic A-type K+ current in CA1. This hypothesis predicts an absence of A-type K+ current in CA1 dendrites of Kv4.2−/− animals.
With dendritic cell-attached voltage-clamp recordings we found that the A-type component of dendritic K+ currents in Kv4.2−/− mice was essentially absent, whereas other components of voltage-dependent K+ currents in the dendrites were intact. The elimination of A-type K+ current was complete in the dendrite, but not in the whole cell. Upregulation of Kv1-mediated channels may compensate for the loss of Kv4.2 in parts of the cell other than the apical dendrite (Fig. 2). A small but significant upregulation of Kv1.4 protein was demonstrated in ventricular cardiac myocytes of Kv4.2−/− mice (Guo et al., 2005). We did not detect a significant upregulation of Kv1.4 protein in the hippocampus with Western blotting. However, this upregulation may be occurring in CA1 pyramidal neurons but may be harder to detect with Western blotting against a background of Kv1.4 expression in other hippocampal cell types. In addition, although an upregulation of Kv1 proteins may compensate for the loss of whole-cell A-type current, the ability of the neuron to compensate may be limited by the fact that it apparently cannot send Kv1 proteins to the dendrites (Gu et al., 2003). Although our data support dendritic A-type current being composed of Kv4.2, we cannot exclude the alternative possibility that downregulation of proteins other than Kv4.2 (KChIPs, for example) (Fig. 1) may have caused concomitant downregulation of channel proteins other than Kv4.2. However, our Western blotting data demonstrate no downregulation of Kv4.3, the principal alternative candidate molecule that might encode dendritic A-type current. It also would be of great interest to explore whether meaningful functional links exist between the downregulation of Kv4.2 and KChIPs (Menegola and Trimmer, 2006). Nevertheless, we believe the most straightforward interpretation of our data is that Kv4.2 is a necessary α-subunit for the dendritic A-type K+ current in CA1.
The specific and complete elimination of A-type K+ current in the CA1 dendrites of Kv4.2−/− mice provided the basis for our interpretation of subsequent experiments. We demonstrated that APs in CA1 pyramidal neurons backpropagate into the dendrites with a decrement in amplitude. This decrement was reduced in Kv4.2−/− dendrites, but was not completely absent. In other words, complete elimination of A-type K+ current in dendrites did not fully restore bAPs to the same amplitude of somatic APs. This observation suggests that other mechanisms also are involved in regulation of bAP, such as dendritic morphology and geometry (Williams and Stuart, 2000; London and Hausser, 2005).
The bigger-amplitude bAPs were accompanied by Ca2+ influx, which decreased much less in the dendrites of Kv4.2−/− animals than in the dendrites of wild-type animals. As a consequence, induction of Hebbian-type LTP was easier in Kv4.2−/− neurons because fewer postsynaptic APs were required to induce robust LTP. Previous studies with selective blockers of Kv4 channels using field recordings reported qualitatively similar results (Ramakers and Storm, 2002). In our studies the maximum amount of LTP was not significantly different between Kv4.2−/− and wild-type animals. It is interesting to consider that the induction of long-term depression (LTD), another form of neural plasticity, also depends on the amount of Ca2+ influx. It remains to be tested whether larger bAPs propagating in the dendrites of Kv4.2−/− also exert their influence on LTD induction.
Given the results of previous studies (Hoffman et al., 1997; Bernard et al., 2004), it came as a surprise that the Kv4.2−/− animals did not have innate epilepsy with a behavioral manifestation. We think perhaps the upregulation of the Kv1-mediated current could have played a compensatory role to counter the expected increase in whole-cell excitability in Kv4.2−/−. However, it remains to be seen whether epileptic forms of activity can be recorded with EEGs in the Kv4.2−/− animals and whether the Kv4.2−/− animals are more prone to epilepsy when challenged with exogenous insults.
In conclusion, the Kv4.2−/− mice have proven to be a useful model for establishing a causal link between a dendritic conductance (A-type K+ current), aspects of neuronal signaling and plasticity (London and Hausser, 2005), and a specific molecule, Kv4.2. Together with previous research (Hoffman et al., 1997; Hoffman and Johnston, 1998; Ramakers and Storm, 2002; Watanabe et al., 2002; Yuan et al., 2002; Frick et al., 2004), our data provide strong support for A-type K+ current determining action potential backpropagation in the dendrites and regulating the induction threshold of synaptic plasticity. However, the A-type K+ current is not the complete story, as we show in the case of action potential backpropagation, and additional experiments are necessary to delineate fully the mechanisms regulating signal propagation in dendrites.
Footnotes
-
This work was supported by National Institutes of Health Grants NS37444, MH48432, MH44754 (D.J.), MH70857 (L.-L.Y.), and MH57014 (J.D.S.). We thank Dr. Andreas Jeromin, Yunqing Han, and Amber Lockridge for genotyping and maintaining the mouse colonies and Dr. James Trimmer for the gift of some antibodies.
- Correspondence should be addressed to Daniel Johnston, Center for Learning and Memory, University of Texas at Austin, Austin, TX 78712. djohnston{at}mail.clm.utexas.edu
