Frequency-dependent regulation of rat hippocampal somato-dendritic excitability by the K⁺ channel subunit Kv2.1

Abstract

1. The voltage-dependent potassium channel subunit Kv2.1 is widely expressed throughout the mammalian CNS and is clustered primarily on the somata and proximal dendrites, but not axons, of both principal neurones and inhibitory interneurones of the cortex and hippocampus. This expression pattern suggests that Kv2.1-containing channels may play a role in the regulation of pyramidal neurone excitability. To test this hypothesis and to determine the functional role of Kv2.1-containing channels, cultured hippocampal slices were incubated with antisense oligonucleotides directed against Kv2.1 mRNA.

2. Western blot analysis demonstrated that Kv2.1 protein content of cultured slices decreased > 90 % following 2 weeks of treatment with antisense oligonucleotides, when compared with either control missense-treated or untreated cultures. Similarly, Kv2.1 immunostaining was selectively decreased in antisense-treated cultures.

3. Sustained outward potassium currents, recorded in both whole-cell and outside-out patch configurations, demonstrated a selective reduction of amplitude only in antisense-treated CA1 pyramidal neurones.

4. Under current-clamp conditions, action potential durations were identical in antisense-treated, control missense-treated and untreated slices when initiated by low frequency stimulation (0.2 Hz). In contrast, spike repolarization was progressively prolonged during higher frequencies of stimulation (1 Hz) only in cells from antisense-treated slices. Similarly, action potentials recorded during electrographic interictal activity in the ‘high [K⁺]_o’ model of epilepsy demonstrated pronounced broadening of their late phase only in cells from antisense-treated slices.

5. Consistent with the frequency-dependent spike broadening, calcium imaging experiments from single CA1 pyramidal neurones revealed that high frequency Schaffer collateral stimulation resulted in a prolonged elevation of dendritic [Ca²⁺]_i transients only in antisense-treated neurones.

6. These studies demonstrate that channels containing Kv2.1 play a role in regulating pyramidal neurone somato-dendritic excitability primarily during episodes of high frequency synaptic transmission.

Voltage-dependent potassium channels play a fundamental role in determining neuronal excitability (Hille, 1991). The electrophysiological properties of most K⁺ channel family members have been studied primarily in heterologous expression systems (e.g. Xenopus oocytes, human embryonic kidney (HEK293) cells). In these expression systems currents through the Shab subfamily member Kv2.1 are activated at low thresholds and are essentially non-inactivating during episodes of prolonged depolarization (Shi et al. 1994). Kv2.1 expression is widespread throughout the mammalian CNS and in the hippocampus is clustered on the soma and dendrites of both principal neurones and interneurones (Hwang et al. 1993; Du et al. 1998). Recent evidence suggests that Kv2.1 is a major component of the ‘delayed rectifier’ recorded from hippocampal neurones (Murakoshi & Trimmer, 1999), although the precise role played by the Kv2.1 conductance has been difficult to demonstrate given the lack of availability of selective antagonists. Somatic or dendritic K⁺ conductances play a major role in regulating the back-propagation of action potentials from the axonal to the dendritic compartment (Johnston et al. 1996; Stuart et al. 1997). The somato-dendritic expression of Kv2.1 suggests that it may be one type of voltage-gated K⁺ channel important for regulation of pyramidal neurone excitability.

Currents through Kv2.1 channels can be blocked by low concentrations of the poorly selective antagonists tetraethylammonium (TEA) and 4-aminopyridine (4-AP) and the recently identified toxin from the Chilean tarantula, hanatoxin (Swartz & MacKinnon, 1995). This toxin, however, also blocks currents through Kv4.2, a subunit with an expression pattern in hippocampal neurones similar to that of Kv2.1, complicating interpretation of the role of Kv2.1 in these cells. Due to the lack of selective antagonists for Kv2.1 channels, we have devised an antisense oligonucleotide strategy to selectively reduce Kv2.1 expression in hippocampal slice cultures (Gahwiler et al. 1997). Using antisense oligonucleotides, Kv2.1 protein expression was decreased by > 90 % as determined by Western blot analysis. Sustained outward currents recorded from antisense-treated CA1 pyramidal neurones were significantly reduced compared to untreated and missense oligonucleotide-treated controls. Furthermore, we determined a role for Kv2.1 in shaping the action potential and the associated intradendritic [Ca²⁺]_i during high frequency synaptic stimulation.

METHODS

Hippocampal cultures

Hippocampal slices (500 μm thick) and cultures were prepared from postnatal day 7 (P7) Sprague-Dawley rats as described previously (McBain et al. 1989; Stoppini et al. 1991). Animals were anaesthetized by inhalation of isoflurane and decapitated according to NIH guidelines. Cultures were incubated at 37°C in 5 % CO₂ in medium of the following composition: 50 % minimal essential medium (MEM) with Eagle's salts, 25 % Hanks’ balanced salt solution (HBSS), 25 % heat-inactivated horse serum, 0.01 M Hepes, 0.5 % GlutaMaxII and 0.65 % glucose, pH 7.2. After 5 days, cultures were treated with a combination of two unique phosphorothioate-modified antisense oligonucleotide sequences (4.5 μM each; sequence I, 5′-G CTT CGT {"Underbar" on}CAT{"Underbar" off} GCC AGA CCA G-3′, corresponding to nucleotides 12–31 of Frech et al. 1989– the translation initiation codon is underlined; sequence II, 5′-TCC TCC CCC AGG CAG CAT GT-3′, nucleotides 2531–2550), or with two control missense sequences (sequence I, 5′-GGC CCC TAA GAC CCT GGT AT-3′; and sequence II, 5′-GGC CCC TAT GAG CCC TTC CA-3′) for periods of up to 2 weeks in vitro. Fresh oligonucleotides were added every other day.

Western blot analysis

Cultured hippocampal slices (10–16 per treatment) were homogenized in lysis buffer after 5 or 14 days of treatment (50 mM Hepes, 1 % Triton X-100, 50 mM NaCl, 50 mM NaF, 10 mM sodium phosphate, 5 mM EDTA, 1 mM sodium orthovanadate, 1 mM phenylmethanesulfonyl fluoride (PMSF), 1 μM pepstatin A and 10 μg ml⁻¹ leupeptin) at 4°C. The homogenates were centrifuged at 14 000 g for 5 min and the supernatants collected. Membrane protein samples (20 μg) were dissolved in sample buffer (100 mM Tris pH 7.6, 1 % SDS, 10 % glycerol, 0.05 % Bromophenol Blue and 5 %β-mercaptoethanol), denatured and separated by 8 % SDS-PAGE. Proteins were transferred onto nitrocellulose membranes and probed with a polyclonal anti-Kv2.1 antibody (1:100; Upstate Biotechnology, Lake Placid, NY, USA), anti-Kv1.5 antibody (1:1000; Upstate Biotechnology) or anti-neurone-specific enolase antibody (1:500; ICN Biomedicals, OH, USA) overnight. Enhanced chemiluminescence (ECL) was used as a secondary detection system.

Kv2.1 immunocytochemistry

Fixed hippocampal cultures were incubated in blocking solution (10 % normal goat serum and 0.2 % Triton X-100 in PBS) for 30 min. The anti-Kv2.1 antibody (1:10) was diluted in 5 % normal goat serum in PBS and incubated with the cells overnight. Either Cy3-conjugated or biotinylated goat anti-rabbit anti-serum was used as a secondary antibody (1:300 and 1:200, respectively).

Electrophysiology

Whole-cell and outside-out patch-clamp recordings were made from visually identified CA1 pyramidal neurones in cultures treated for 2 weeks with oligonucleotides or from parallel untreated cultures. Cultures were perfused with the following solution (mM): 130 NaCl, 24 NaHCO₃, 3.5 KCl, 1.25 NaH₂PO₄, 1.0 CaCl₂, 1.2 MgSO₄, 10 glucose and 0.001 TTX, saturated with 95 % O₂-5 % CO₂, pH 7.4. To eliminate activation of Ca²⁺-activated potassium currents, 0.2 mM Cd²⁺ was added to the recording solution. For current-clamp recordings, both TTX and Cd²⁺ were omitted from the recording solution. Patch electrodes had resistances of 2–6 MΩ when filled with the following solution (mM): 130 potassium gluconate, 10 NaCl, 2 Na₂-ATP, 0.3 Na-GTP, 10 Hepes and 0.6 EGTA, with 0.4 % biocytin, pH 7.4 and ∼275 mosmol l⁻¹. Linear leak current and capacitive artifacts were digitally subtracted on-line using a -P/4 or -P/8 routine. All data are expressed as means ±s.e.m.

Calcium imaging measurements

Patch pipettes were filled with (mM): 140 potassium methanesulfonic acid, 10 Hepes, 4.0 Na₂-ATP and 0.6 Na-GTP, containing the cell-impermeant fluorescent calcium indicator Calcium Green-1 hexapotassium salt (100 μM; Molecular Probes). Cells were whole-cell voltage clamped (holding potential (V_h) =−60 mV) for a period of ∼5 min to allow adequate filling with fluorescent indicator. Pyramidal cells were imaged using a confocal microscope coupled with an Odyssey II scan unit (Noran, Middleton, WI, USA) using a standard fluorescein filter set. The Schaffer collateral pathway was stimulated 5 times at 0.03 or 1 Hz using a bipolar electrode. Images were acquired at 60 frames s⁻¹. Fluorescence intensity was extracted using Intervision 2D software (Noran) and normalized to pre-stimulation levels (ΔF/F₀).

RESULTS

Immunohistochemical localization of Kv2.1

Kv2.1 was widely expressed throughout the hippocampal formation including the pyramidal cell layers, inhibitory interneurones, the dentate gyrus and the hilar region (Fig. 1) consistent with previous reports (Trimmer, 1991; Scannevin et al. 1996; Du et al. 1998). At the light microscopic level, Kv2.1 expression was clustered on the soma and proximal dendrites but was absent from axons of virtually all neurones, suggesting that it may play a significant role in shaping dendritic excitability.

Figure 1. Kv2.1 has a somato-dendritic expression in the hippocampus.

A, Kv2.1 protein was expressed in virtually every cell from all subfields of the hippocampal formation (× 6.5 magnification). B and C, high power photomicrographs of the CA1 subfield and the dentate gyrus/hilar region, respectively. Kv2.1 expression was localized to the soma and proximal dendrites of both principal neurones (pyramidal cells and granule cells) as well as inhibitory interneurones located in the stratum oriens, radiatum and hilus (× 16.25 magnification).

Antisense oligonucleotides decrease Kv2.1 protein expression

Two unique phosphorothionate-modified antisense oligonucleotides were designed, one complementary to the proximal translation initiation site and the other to prevent translation elongation of Kv2.1. As a control, parallel cultures were exposed to two missense oligonucleotide sequences, composed of scrambled antisense oligonucleotide sequences, or alternatively, received no treatment (‘untreated’). In initial experiments we 5′ end-labelled the oligonucleotides with a rhodamine fluorescent tag (6-FAM dye phosphoamadite) to determine the time course of their uptake into hippocampal neurones. After several hours of exposure numerous cells were observed to contain the oligonucleotides and after ∼3 days of treatment, confocal microscopy revealed that oligonucleotides were present in virtually every cell of the hippocampal slice (Fig. 2). The pattern of oligonucleotide uptake was not altered by 7 days suggesting that maximal uptake had occurred by 3 days in vitro.

Figure 2. The time course of oligonucleotide uptake into pyramidal neurones.

To determine the time course of oligonucleotide uptake into hippocampal pyramidal neurones initial experiments were made using oligonucleotides 5′ end-labelled with a rhodamine fluorescent tag (6-FAM dye phosphoamadite). Confocal microscopy revealed that after 3 days virtually all cells contained visible levels of oligonucleotide within the cytoplasmic compartment (A). This level of uptake was not altered after 7 days despite fresh oligonucleotide being added every second day (B). Magnification, × 32.

Kv2.1 protein expression was quantified by Western blot analysis in all three conditions (Fig. 3). An antibody selective for Kv2.1 recognized two bands on a Western blot at ∼110-130 kDa, consistent with previous reports (Shi et al. 1994). The inclusion of the antisense oligonucleotides in the culture media for 5 days only slightly reduced Kv2.1 protein expression (69.8 ± 7.5 %, n = 3 individual experiments) compared to parallel untreated control slices. Kv2.1 expression in missense-treated cultures was 85.3 ± 5.6 % of untreated slices (n = 3, Fig. 3a). However, following 2 weeks of antisense oligonucleotide treatment, Kv2.1 expression was greatly reduced to 7.8 ± 5.7 % of untreated slices (n = 6). At this time point, there was a high level of Kv2.1 expression in missense-treated cultures (72.4 ± 22 % of untreated, Fig. 3b). As a further control, we also monitored the level of Kv1.5, a subunit expressed in high levels in both neurones and glia of the hippocampus (Maletic-Savatic et al. 1995; Roy et al. 1996). A polyclonal antibody against Kv1.5 recognized a single band on a Western blot of ∼76 kDa. Treatment of slices with either antisense or control missense Kv2.1 oligonucleotides was without effect on Kv1.5 expression suggesting that oligonucleotide treatment did not result in a non-specific knockdown of expression of other K⁺ channel subunits (Fig. 3C).

Figure 3. Antisense oligonucleotides reduce Kv2.1 protein expression in hippocampal cultures.

A and B, Western blot analysis of membrane protein derived from hippocampal organotypic cultures treated for 5 and 14 days with a combination of two antisense oligonucleotides (AS, Antisense), two missense oligonucleotides (C, Control), or receiving no oligonucleotide treatment (UT, Untreated). A polyclonal antibody selective for Kv2.1 recognized two bands on a Western blot at ∼110-130 kDa. A, at 5 days, only antisense-treated cultures showed a significant reduction in Kv2.1 expression: antisense, 69.8 ± 7.5 %; missense, 85.3 ± 5.6 % of untreated (n = 3). B, following 14 days of antisense treatment, Kv2.1 expression was greatly reduced (7.8 ± 5.7 % of untreated). At the same time point, Kv2.1 expression was only reduced to 72.4 ± 22 % of untreated in missense-treated cultures (n = 6). C, as a control, the same samples were probed with a polyclonal antibody to Kv1.5. The Kv1.5 antibody recognized a single band on the Western blot of ≈76 kDa. Note that there was no significant change in Kv1.5 protein expression in either Kv2.1 antisense- or missense-treated cultures compared to untreated cultures, demonstrating that neither form of Kv2.1 oligonucleotide reduces expression of an unrelated K⁺ channel subunit. D, as a further control we monitored the expression of neurone-specific enolase, a specific neuronal marker, to determine whether the reduction of Kv2.1 in antisense-treated cultures resulted from non-specific neuronal death. A polyclonal antibody raised against neurone-specific enolase recognized a single band on the Western blot of ∼90 kDa. Expression of neurone-specific enolase in both antisense-treated and control missense-treated conditions was not significantly different from that in untreated cultures (data from 3 independent culture preparations). These data demonstrate that the reduction of Kv2.1 expression observed under antisense treatment does not arise from a non-specific loss of neurones from the culture.

We also monitored the level of neurone-specific enolase, a specific neuronal marker, to determine whether the reduction of Kv2.1 in antisense-treated cultures resulted from non-specific neuronal death. A polyclonal antibody raised against neurone-specific enolase (ICN Biomedicals) recognized a single band on the Western blot of ∼90 kDa according to manufacturer's guidelines. Expression of neurone-specific enolase in antisense-treated and control missense-treated conditions was 97.4 ± 0.01 and 97.7 ± 0.06 %, respectively, of the untreated control (Fig. 3D, data from three independent culture preparations). These data demonstrate that the reduction of Kv2.1 expression observed under antisense treatment does not arise from a non-specific loss of neurones from the culture.

Neither the gross morphology of cultured hippocampal slices nor the morphology of individual neurones was altered by any of the treatment conditions throughout the treatment period. Figure 4A–C shows typical examples of Nissl-stained cultured slices (14 days in vitro) from all three conditions. Furthermore all individual neurones used for electrophysiological recordings (see below for details) were filled with biocytin and their anatomy recovered post hoc. Figure 4D–F shows typical pyramidal neurones from all three culture conditions demonstrating that individual cell morphology was also unaltered during the oligonucleotide treatment.

Figure 4. Neither gross hippocampal anatomy nor pyramidal neurone morphology is altered following reduction of Kv2.1 protein expression.

A–C, Nissl staining of representative antisense-treated, control missense-treated and untreated cultures reveals that all major subfields of the hippocampal formation remained unaltered following 2 weeks of oligonucleotide treatment (× 2.6 magnification). D–F, similarly individual pyramidal neurones filled with biocytin during whole-cell electrophysiological recordings showed normal dendritic arbors in all three oligonucleotide treatment conditions (× 6.5 magnification). G–I, to further confirm that Kv2.1 expression was selectively reduced following antisense oligonucleotide treatment, we performed Kv2.1 immunohistochemistry on cultures maintained in parallel with those used for Western blot analysis (shown in Fig. 3) (× 26 magnification). Using confocal microscopy no differences in the expression pattern were observed between control missense-treated (H) and untreated cultures (I). Both images were obtained from the CA1 pyramidal cell layer using identical acquisition parameters and laser intensity (30 %). In contrast, sections from antisense-treated cultures (G) showed little, if any, Kv2.1 protein expression. The antisense image was obtained using the same acquisition parameters except that the laser intensity was set at 100 %.

To further confirm that antisense oligonucleotide treatment reduced the Kv2.1 expression level we next performed immunohistochemistry on cultures treated for 2 weeks. Using confocal microscopy we demonstrated that Kv2.1 expression was reduced only in antisense-treated cultures (Fig. 4G–I). Kv2.1 expression was localized to the soma and proximal dendrites of pyramidal neurones in missense-treated and untreated cultures but was largely absent from antisense-treated cells (n = 6 independent preparations).

Antisense oligonucleotides reduce the sustained current in pyramidal neurones

Consistent with the reduction in the Kv2.1 protein content of hippocampal neurones, electrophysiological studies revealed that the amplitude of sustained outward potassium currents was selectively reduced in both the whole-cell configuration and outside-out patches excised from antisense-treated CA1 pyramidal neurones. Outward currents were evoked by a step to +40 mV (V_h =−40 mV, 10 mV increments) to selectively activate sustained currents (Zhang & McBain, 1995) and were measured at a time point of 400 ms. At a test potential of +40 mV sustained currents were reduced by 56 % (2046 ± 239 versus 4622 ± 414 pA in untreated; Student's unpaired t test, P < 0.0001; n = 22) in whole-cell recordings and 62 % (103 ± 16 versus 270 ± 88 pA in untreated; Student's unpaired t test, P = 0.03; n = 18) for outside-out patch recordings (Fig. 5C and D). In contrast, sustained currents from missense-treated cultures were not significantly different from those of untreated cells in both whole-cell (P = 0.15, n = 9) and outside-out patches (P = 0.96, n = 9) (Fig. 5C and D). Despite the marked reduction of sustained current in antisense-treated cells, the voltage dependence of sustained currents remaining in antisense-treated cells was not significantly different from that of currents in untreated cultures (V_½= 11.0 ± 0.2 mV and k = 10.9 ± 0.2 versusV_½= 4.1 ± 0.8 mV and k = 13.3 ± 1.2, respectively, n = 10; Fig. 5b). However, an ∼15 mV positive shift in the threshold for sustained current activation (-17 ± 2.2 mV, n = 20, versus−31 ± 2.1 mV in untreated, n = 23) was observed in antisense-treated cultures compared to control untreated cultures.

Figure 5. Sustained outward currents are reduced in Kv2.1 antisense-treated cells.

Electrophysiological recordings demonstrate that the reduction of Kv2.1 protein expression by antisense oligonucleotides reduces the sustained outward current in CA1 pyramidal neurones. A, outside-out macropatches were pulled from the soma of CA1 pyramidal neurones from cultures receiving either antisense or missense oligonucleotide treatment for 14 days or from untreated cultures. Sustained outward currents were activated by test pulses from a holding potential of −40 mV to +70 mV (10 mV increments). The overall kinetic and voltage-dependent properties of the currents in all three conditions were similar. B, the activation threshold of currents from antisense-treated pyramidal neurones was shifted from a mean of −31 ± 2.1 mV (n = 23) in untreated neurones to −17 ± 2.2 mV (n = 20). In contrast, the half-activation values of currents from untreated and antisense-treated cells were identical. The voltage dependence of activation is shown for 10 representative outside-out patches from both the untreated (▪) and antisense-treated (•) condition. The mean half-activation potential (V_½) was 11.0 ± 0.2 mV and k = 10.9 ± 0.2 in antisense-treated cultures versusV_½= 4.1 ± 0.8 mV, k = 13.3 ± 1.2 in untreated cultures. C and D, sustained outward currents in both whole-cell (C) and outside-out patches (D) were significantly reduced (** P < 0.05) only in CA1 pyramidal neurones from antisense-treated cultures when compared to either missense-treated or untreated cultures. Current amplitudes were measured at a test pulse to +40 mV (V_h=−40 mV), at a time point of 400 ms. The number of experiments is given in parentheses. These data demonstrate that channels formed from Kv2.1 subunits are partially responsible for sustained outward current in CA1 pyramidal neurones.

Knockdown of Kv2.1 affects high but not low frequency action potential activity

In current-clamp recordings, resting membrane parameters were identical between all three groups of cells. The mean resting membrane potentials and input resistances were −55.6 ± 3.5 mV and 158.1 ± 37.1 MΩ for antisense-treated, −57.2 ± 4.0 mV and 192.2 ± 21.7 MΩ for missense-treated, and −62.8 ± 2.4 mV and 154 ± 11.3 MΩ for untreated CA1 pyramidal neurones (n = 15, 7 and 11, respectively). Action potential waveforms evoked by somatic depolarization (V_h=−60 mV, 10–100 pA current pulses) were not significantly different in antisense-treated cells (duration at half-amplitude, 3.5 ± 0.5 ms; duration, 6.8 ± 1.2 ms; amplitude, 77.3 ± 5.8 mV; n = 9), when compared to either control missense-treated (duration at half-amplitude, 2.7 ± 0.3 ms; duration, 4.2 ± 0.5 ms; amplitude, 100.8 ± 9.1 mV; n = 8) or untreated cells (duration at half-amplitude, 2.6 ± 0.1 ms; duration, 4.6 ± 0.3 ms; amplitude, 97.8 ± 10.1 mV, n = 7) recorded after 14 days of treatment (Fig. 6a and B). These data would suggest that the K⁺- and Ca²⁺-activated K⁺ conductances (Halliwell, 1990) remaining within antisense-treated cells are sufficient to repolarize the action potential waveform even in the absence of a significant Kv2.1 conductance.

Figure 6. Action potential activity is selectively broadened during electrographic interictal activity in antisense-treated cultures.

Despite a large reduction in sustained outward current, action potential activity was unaltered under conditions of normal [K⁺]_o in antisense-treated cultures (A) compared to control missense-treated cultures (B). Left panels, current injection (100 pA increments) revealed no differences in the current-voltage relationship between antisense- and missense-treated CA1 pyramidal neurones. Middle panels, similarly single spontaneous action potentials were not different in the two culture conditions. Right panels, in contrast upon exposure to an elevated [K⁺] extracellular solution (8.5 mM), cells in both culture conditions were depolarized by 2–5 mV (dotted line indicates resting potential in 3.5 mM [K⁺]_o). Concomitant with the membrane depolarization, cells fired interictal bursts of action potentials (not shown) interspersed with single action potential firing. A typical spontaneous action potential following exposure to elevated [K⁺]_o solution is shown for each condition. Action potentials in antisense-treated CA1 pyramidal neurones showed a distinct broadening of the late phase of action potential repolarization that was not observed in missense-treated pyramidal neurones. Note the change in time scale for the spike in A, right panel.

We considered it possible that the absence of the Kv2.1 conductance within antisense-treated cells may affect action potential activity only during episodes of high frequency activity. To test this hypothesis we examined high frequency action potential firing under two conditions: during interictal bursting in an elevated potassium model of epilepsy and during high frequency synaptic transmission resulting from stimulation of the Schaffer collateral-CA1 synapses.

The high [K⁺]_o model of hypersynchronous activity is produced by elevating the extracellular potassium ion concentration to 8.5 mM (McBain et al. 1993). Consequently interictal bursts, defined as rapid and brief depolarizations coincident with high frequency action potential firing, are observed in the CA1 subfield. The elevation of [K⁺]_o causes a 23 mV positive shift in the reversal potential for potassium ions, resulting in a decreased driving force for K⁺ conductances involved in action potential repolarization. Following the introduction of 8.5 mM [K⁺]_o the pyramidal cell membrane potential was typically depolarised by 2–10 mV. For subsequent analysis of all action potential waveforms the membrane potential was temporarily returned to the control value (-60 mV) by current injection. Under these conditions spontaneous action potentials in both untreated and missense-treated cultures were significantly prolonged (duration at half-amplitude, 3.2 ± 0.4 ms; duration, 14.3 ± 6.4 ms in untreated cultures; n = 3; and duration at half-amplitude, 3.7 ± 0.5 ms; duration, 15.3 ± 5.7 ms in missense-treated cultures; n = 3) compared to events occurring in control conditions of 3.5 mM [K⁺]_o (see above for details). Action potentials in antisense-treated CA1 pyramidal cells were almost 3 times as long as action potentials in untreated and missense-treated slices. In all cells tested a significant broadening of both the duration at half-amplitude (6.9 ± 1.5 ms) and the late component of the action potential waveform was observed (mean duration, 48.6 ± 14.2 ms; n = 5; Fig. 6). On several occasions the late phase of the action potential observed in 8.5 mM [K⁺]_o was accompanied by a large plateau potential, which presumably occurred as a result of calcium entry during the late phase of the action potential. These data demonstrate that the Kv2.1 conductance is essential for complete action potential repolarization during interictal activity.

Under current-clamp conditions, action potential waveforms activated from suprathreshold EPSPs, evoked by stimulation of the Schaffer collateral-commisural pathway, were also frequency dependent in antisense-treated cells. At a low stimulation frequency (0.2 Hz), action potentials occurring early in the train were identical to those observed late in the train (Fig. 7b). In contrast, at higher frequencies (1 Hz) the late repolarizing phase of each action potential broadened during subsequent spikes. The spike broadening was cumulative until complete repolarization was no longer possible and a plateau depolarization was observed (n = 4). In both missense-treated and untreated neurones, action potential waveforms were identical throughout the 1 Hz train (n = 5 and 4, respectively; Fig. 7a and C). In control untreated cells exposed to 10 mM TEA, which blocks various delayed rectifier currents in hippocampal neurones (Zhang & McBain, 1995), the late phase of action potential repolarization was broadened similar to antisense-treated cells (Fig. 7A, n = 3). These data suggest that at high but not low frequencies of synaptic transmission, successful repolarization of action potentials is highly dependent on an active Kv2.1 conductance.

Figure 7. Action potential broadening during high frequency synaptic stimulation.

A–C, suprathreshold stimulation of the Schaffer collateral pathway activated an EPSP-action potential sequence in CA1 pyramidal neurones from untreated (A, Control), antisense-treated (B) and missense-treated slices (C) (all cultures treated for 14 days). EPSPs and action potentials evoked at 0.2 Hz were superimposable in all three conditions. At 1 Hz stimulation frequency, EPSPs and action potentials in both untreated and missense-treated cultures were similarly superimposable. In contrast, the late phase of the action potential in antisense-treated CA1 pyramidal neurones showed a progressive broadening until on the fifth stimulus the action potential duration was in excess of 200 ms (complete repolarization truncated for clarity). A similar broadening of the late phase of the action potential was observed in untreated cultures exposed to TEA (10 mM), an antagonist of currents through Kv2.1 channels.

Since the expression of Kv2.1 is highest in the soma and dendrites of hippocampal principal cells, we wanted to determine whether the frequency-dependent prolongation of action potentials observed in the previous set of experiments affected dendritic calcium signalling. Intracellular Ca²⁺ responses generated by stimulation of the Schaffer collateral inputs were studied in single Calcium Green-1-filled CA1 pyramidal cells. Figure 8A–C shows the normalized [Ca²⁺]_i responses from the proximal apical dendrite region of three representative cells from untreated control, antisense- and missense-treated cultures. Low frequency stimulation of the Schaffer collaterals evoked a calcium transient with a decay best described by two exponentials. In untreated cells the time constants of decay were 269 ± 45 ms and 1.6 ± 0.2 s (n = 3). In antisense-treated cells the initial time constant was identical but the second component was prolonged compared to both untreated and missense-treated cells (τ₁ = 293 ± 54 ms and τ₂ = 3.9 ± 0.6 s; Student's unpaired t test, P = 0.028; n = 3). Missense-treated cells had [Ca²⁺]_i transients similar to those of untreated cells (τ₁ = 286 ± 35 ms and τ₂= 1.3 ± 0.4 s, n = 3). At 1 Hz (5 stimuli), a frequency which dramatically altered the action potential waveform in antisense-treated cells (Fig. 8b), [Ca²⁺]_i transients were significantly prolonged in antisense-treated cells compared to control untreated or missense-treated cells (Fig. 8A–C). In antisense-treated cells, the resulting [Ca²⁺]_i transient took upwards of 5–10 s to return to the baseline value compared to 2–4 s for untreated or missense-treated cultures. Furthermore, in antisense-treated neurones, the [Ca²⁺]_i transient was observed to continue to rise following the cessation of Schaffer collateral stimulation consistent with the prolonged plateau potentials seen in current-clamp recordings of action potentials in these cells (Fig. 7). Identical trends were observed when the [Ca²⁺]_i signal was monitored in the soma and basilar dendritic tree (data not shown).

Figure 8. Frequency-dependent prolongation of evoked [Ca²⁺]_i transients in cells from antisense-treated cultures.

A–C, intracellular Ca²⁺ responses were studied in Calcium Green-1-filled CA1 pyramidal cells treated with Kv2.1 antisense (B) or missense oligonucleotides (C) for 14 days, or untreated (A, Control). Shown are normalized [Ca²⁺]_i responses from the proximal apical dendrite region of three representative cells in response to Schaffer collateral stimulation. Left panels, [Ca²⁺]_i responses from five trials at 0.03 Hz, normalized and averaged. The decay time constant (τ) was determined by fitting this averaged response with exponential curves. Responses were fitted best with two exponential curves. τ₁, the rapid initial component, was similar in CA1 pyramidal cells in antisense-treated (τ₁= 293 ± 54 ms), missense-treated (286 ± 36 ms) and untreated (269 ± 45 ms) slices. However, the second decay time constant was significantly prolonged in pyramidal cells in antisense-treated slices (3.9 ± 0.6 s compared to 1.3 ± 0.4 s in missense-treated and 1.6 ± 0.2 s in untreated slices, P = 0.03). Right panels, [Ca²⁺]_i signal in response to 1 Hz stimulation (5 stimuli). Note that the increased [Ca²⁺]_i transient in the antisense-treated pyramidal neurone was significantly prolonged compared to the [Ca²⁺]_i transient in both untreated and missense-treated cells.

DISCUSSION

Antisense knockdown has been used previously with considerable success against a variety of K⁺ channel subunits both in vitro (Chung et al. 1995; Roy et al. 1996) and in vivo (Meiri et al. 1997, 1998). Here the use of antisense oligonucleotides allowed us to perform experiments that would otherwise be impossible due to the lack of selective antagonists to Kv2.1 channels. In the present study we found that Kv2.1 expression determined by Western blot and immunohistochemical analysis was reduced to < 10 % of control following 14 days of antisense oligonucleotide exposure. Subsequent electrophysiological recordings from CA1 pyramidal neurones revealed that sustained outward currents were selectively reduced in recordings made in both the whole-cell configuration and outside-out patches excised from the cell soma or proximal dendrites. Of particular interest, we found that under conditions of low frequency stimulation, an absence of Kv2.1 did not result in a significant perturbation in the action potential waveform; repolarization was complete and the duration was similar to that in untreated or missense-treated cells. This results presumably from the ability of the remaining K⁺ conductances to fully repolarize the action potential waveform during periods of low frequency stimulation. Under higher stimulation frequencies both spike duration and dendritic calcium transients were progressively prolonged suggesting that Kv2.1 is essential for repolarization of action potentials during high frequency stimulation. Similar results were observed during high frequency interictal firing in an elevated [K⁺]_o model of epilepsy. Spike broadening may arise as a consequence of the remaining K⁺ conductances accumulating in an inactivated state or being unable to deinactivate sufficiently to follow the action potential waveform. Alternatively, it is possible that upregulation of voltage-gated calcium channel expression may accompany the antisense knockdown of Kv2.1, allowing an increased entry of calcium during high frequency activation. This latter hypothesis is unlikely given the lack of effect of low frequency stimulation on both spike width and the dendritic calcium transients in antisense-treated cells.

These data provide a direct demonstration that a sustained current carried by Kv2.1 channels plays a role in regulating high frequency action potential activity and consequently regulates dendritic calcium entry in CA1 pyramidal neurones.

Previous studies have demonstrated that Kv2.1 expression is widespread throughout the mammalian CNS (Trimmer, 1991; Hwang et al. 1993; Maletic-Savatic et al. 1995; Scannevin et al. 1996; Du et al. 1998). At the electron microscopic level Kv2.1 expression was clustered on neuronal cell bodies and dendrites of both principal neurones and somatostatin-, calbindin- and parvalbumin-positive inhibitory interneurones (Du et al. 1998). In all cell types so far studied, Kv2.1 was absent from axonal processes (both myelinated and unmyelinated) suggesting that channels containing Kv2.1 may be candidates for a role in the regulation of pyramidal cell somato-dendritic excitability. Currents with properties resembling those of Kv2.1 currents have been identified in hippocampal CA1 pyramidal neurones and a variety of inhibitory interneurones (Zhang & McBain, 1995; Chikwendu & McBain, 1996; Martina et al. 1998). RT-PCR techniques have demonstrated the presence of Kv2.1 mRNA in > 60 % of CA1 pyramidal neurones and ∼20 % of dentate gyrus basket cells tested (Martina et al. 1998). Recently, Murakoshi & Trimmer (1999) demonstrated, using a Kv2.1 antibody as a functional antagonist, that current through Kv2.1 channels makes a significant contribution to the delayed rectifier K⁺ current in cultured hippocampal neurones. Our data confirm and extend these previous reports and we now demonstrate an essential role for Kv2.1 conductances in spike repolarization during two types of high frequency activity.

The high level of expression of Kv2.1 in both principal neurones and specific inhibitory neurone classes suggests that currents through these channels may be a common property for most hippocampal neurones. Currents through channels formed from recombinant Kv2.1 subunits expressed in a variety of cell lines possess a sustained ‘delayed-rectifier’ phenotype, which activates at potentials > −20 mV, with a relatively long time to peak amplitude (> 100 ms) and shows little steady-state inactivation (Shi et al. 1994). Recombinant Kv2.1 currents are blocked by low millimolar concentrations of TEA and 4-AP. A comparison of recombinant Kv2.1 current properties with those of sustained currents observed in pyramidal neurones (this study; Numann et al. 1987) and hippocampal interneurones of the stratum oriens-alveus (Zhang & McBain, 1995), stratum pyramidale (Du et al. 1996) and stratum radiatum-lacunosum-moleculare (Chikwendu & McBain, 1996) revealed currents with properties having little in common with recombinant Kv2.1 currents. Although all these cell types possess sustained currents with somewhat similar kinetics, all the currents differ markedly in their sensitivity to TEA and 4-AP, when compared to recombinant Kv2.1 currents. It is highly likely, however, that native channels containing Kv2.1 subunits may be modulated by any number of second messengers, which modify their function. Rat brain Kv2.1 polypeptides are serine phosphorylated extensively within their C-terminal domain (Murakoshi et al. 1997). Phosphorylation of Kv2.1 alters the voltage dependence of activation without altering the macroscopic kinetics. Whether phosphorylation similarly alters the ability of 4-AP and TEA to block currents through Kv2.1 remains unknown. Alternatively, differences in sustained current properties amongst hippocampal cells expressing Kv2.1 may result from the heteromeric assembly of Kv2.1 with other subunits. Heteromeric assembly with Kv2.2, another Shab subfamily member, is unlikely since its expression within the hippocampal formation is exceptionally low. Recently, however, a new ‘silent’α subunit, Kv2.3, was cloned, which when co-expressed with Kv2.1 subunits resulted in macroscopic K⁺ currents with slowed kinetics and altered voltage dependence (Castellano et al. 1997; Salinas et al. 1997; Chiara et al. 1999). Like Kv2.1, Kv2.3 is widely expressed in principal neurones throughout the hippocampal formation suggesting that Kv2.1-Kv2.3 channel complexes might occur in vivo. Such co-expression in individual neurones may be a mechanism to fine-tune the properties of sustained outward currents in distinct populations of neurones. Experiments involving Kv2.1 or Kv2.3 gene knockout or treatment with antisense oligonucleotides directed at Kv2.3 will be required to fully elucidate the role of Kv2.1 subunits in native channels within particular neuronal populations. Due to a lack of availability of antibodies against Kv2.3 it was impossible in the present experiments to determine whether knockdown of Kv2.1 altered the expression of Kv2.3. However, this would be interesting to pursue in future experiments, designed to investigate Kv2.1 function.

Voltage-gated channels expressed in the dendrites of cortical and hippocampal neurones have been shown to influence dendritic excitability and [Ca²⁺]_i levels (Johnston et al. 1996; Stuart et al. 1997). In hippocampal pyramidal neurones, transient A-type potassium currents, putatively identified as Kv4.2 currents, regulate a variety of dendritic processes (Hoffman et al. 1997). These channels prevent action potential initiation in the dendrites, limit spike back-propagation into the dendrites and reduce excitatory synaptic events. Transient A-type channels increase in density with distance from the soma while sustained outward currents were uniformly expressed over the proximal and distal dendrites. Despite overwhelming evidence for a role for transient currents in regulating dendritic excitability little evidence exists for a major role for sustained currents within pyramidal neurone dendrites.

Previous computer simulations have suggested that the relatively slowly activating sustained current observed in the soma and dendrites of pyramidal neurones would have little role to play in the regulation of the back-propagating action potential (Hoffman et al. 1997). In the present study we now show that the Kv2.1 conductances shape somatic action potential activity and regulate the properties of dendritic [Ca²⁺]_i transients. Under conditions of low frequency synaptic transmission, action potentials activated by either suprathreshold synaptic stimuli or somatic depolarization are capable of complete repolarization even in cells from cultures demonstrating a significant reduction in Kv2.1 expression. This suggests that Kv2.1 conductances play only a modest role in shaping the action potential waveform, presumably due to the presence of other K⁺ conductances. In contrast, higher stimulation frequencies resulted in a progressive broadening of the action potential until complete repolarization was no longer achieved in antisense-treated cultures and action potentials occurring during interictal activity were similarly broadened. Although all of our electrophysiological recordings were somatic recordings from principal cells, and therefore we cannot directly conclude that Kv2.1 affects dendritic function, evidence that a reduction in Kv2.1 expression can directly influence dendritic excitability comes from our calcium imaging experiments. In these experiments stimulation of Schaffer collaterals at frequencies that influenced action potential activity measured at the soma resulted in a prolongation of the [Ca²⁺]_i transients observed within both the apical and basilar dendrites. These experiments provide direct evidence that the duration of the dendritic [Ca²⁺]_i transient is partially determined by sustained current activity.

Further evidence of a role for sustained currents in regulating dendritic excitability comes from a recent study by Colbert & Pan (1999), who elegantly demonstrated a role for sustained currents in ‘stabilizing’ dendritic excitability following block of the dendritic transient current by arachidonic acid (AA). In these experiments AA blocked the transient outward current while enhancing the availability of the dendritic sustained current. The net effect of AA was to enhance the amplitude of the dendritic action potential without compromising the electrical stability of the dendrites. Of particular interest, the single channel conductance of the dendritic sustained current was close to that reported for recombinant Kv2.1 single channels, suggesting that current through Kv2.1 may have a major role to play in regulating dendritic excitability.

In summary, these data suggest that Kv2.1 expressed on the cell soma and dendrites plays a role in the fidelity of high frequency synaptic transmission and shapes the underlying [Ca²⁺]_i transient, which may play a role in the generation of long term changes in synaptic efficacy. These experiments highlight the important role of potassium channels in the signal processing of hippocampal neurones and confirm the hypothesis that dendritic excitability is controlled by locally expressed voltage-gated potassium channels.