Review
Molecular physiology and modulation of somatodendritic A-type potassium channels

https://doi.org/10.1016/j.mcn.2004.06.011Get rights and content

The somatodendritic subthreshold A-type K+ current (ISA) in nerve cells is a critical component of the ensemble of voltage-gated ionic currents that determine somatodendritic signal integration. The underlying K+ channel belongs to the Shal subfamily of voltage-gated K+ channels. Most Shal channels across the animal kingdom share a high degree of structural conservation, operate in the subthreshold range of membrane potentials, and exhibit relatively fast inactivation and recovery from inactivation. Mammalian Shal K+ channels (Kv4) undergo preferential closed-state inactivation with features that are generally inconsistent with the classical mechanisms of inactivation typical of Shaker K+ channels. Here, we review (1) the physiological and genetic properties of ISA, (2) the molecular mechanisms of Kv4 inactivation and its remodeling by a family of soluble calcium-binding proteins (KChIPs) and a membrane-bound dipeptidase-like protein (DPPX), and (3) the modulation of Kv4 channels by protein phosphorylation.

Introduction

Neuronal signal processing centers on the dendrites, delicate processes that emerge from the somata of most neurons. Whether the signal originates from external sources (such as chemical or mechanical stimuli) or from synaptic terminals, dendrites receive, process, and integrate the respective sensory or synaptic inputs to control the firing of the action potential that is typically initiated at the axon hillock and transmitted down the axon to the nerve terminal.

Far from being passive recipients of input stimuli, dendrites have active properties with important consequences for synaptic integration (Hausser et al., 2000, Johnston et al., 2003, Llinas et al., 1969, Spencer and Kandel, 1961, Yuste and Tank, 1996). Pioneering intracellular recordings using sharp microelectrodes showed the existence of large regenerative dendritic spikes caused by the opening of voltage-gated Na+, Ca2+, and K+ channels (Johnston et al., 1996, Llinas and Nicholson, 1971). This indicates that, in addition to the soma and the axon, action potentials can be generated in the dendrites (Johnston et al., 1996). In many neurons, action potentials initiated in the axon hillock propagate orthodromically down the axon and also into the dendrites in a retrograde manner (backpropagating action potentials) (Stuart et al., 1997). In the dendritic tree, these action potentials serve as signals that report the status of the neuron's output. The generation and backpropagation of action potentials along proximal and distal dendrites can be differentially influenced by the nonuniform patterns of expression of dendritic voltage-gated ion channels (Hausser et al., 2000, Yuste and Tank, 1996). The transient subthreshold A-type K+ current (ISA) in dendrites attenuates the backpropagating action potentials (Hoffman et al., 1997). This attenuation correlates with an increasing density of the ISA with increasing distance from the soma in cerebellar Purkinje cells, CA1 hippocampal pyramidal cells, and cortical pyramidal cells (Hoffman et al., 1997, Johnston et al., 2003, Migliore et al., 1999). Under resting conditions, ISA shunts the action potential as it tries to spread into the distal regions of the dendritic tree. However, when excitatory synaptic inputs and somatic action potentials are paired within a certain time window, the ensuing subthreshold depolarization in distal dendrites inactivates ISA (Johnston et al., 2000, Migliore et al., 1999). Consequently, the attenuation of the backpropagating action potential is substantially reduced. This interaction provides a coincidence detection mechanism that may play an important role in dendritic Ca++ signaling, signal integration, and synaptic plasticity (Hausser et al., 2000, Johnston et al., 2003, Watanabe et al., 2002).

Other aspects of neuronal excitability are also regulated by ISA in mammalian neurons, including adapting spiking behavior of sympathetic cervical neurons (Malin and Nerbonne, 2000), action potential waveform and duration (Bardoni and Belluzzi, 1993), latency to the first spike (Schoppa and Westbrook, 1999, Shibata et al., 2000), and the timing of synaptic inputs (Schoppa and Westbrook, 1999). The physiological roles of ISA depend on its relatively rapid development of inactivation, rapid subthreshold activation, subthreshold steady-state inactivation, and rapid recovery from inactivation at hyperpolarized membrane potentials. The rapidly inactivating neuronal K+ current was first described in seminal voltage-clamp studies of molluscan neurons by Hagiwara et al. (1961), Neher (1971), Connor and Stevens (1971b) (who coined the name “A-type” K+ current IA), and Thompson (1977). Connor and Stevens (1971a) established that the regulation of slow repetitive spike firing was directly related to the kinetic and voltage-dependent properties of the somatic A-type K+ current. Although at steady-state ISA is significantly inactivated at the typical neuronal resting membrane potential (−50 to −60 mV), the ISA channels quickly recover from inactivation during the transient after-hyperpolarization that follows the action potential. Then, as the membrane continues to depolarize, ISA activates in the subthreshold range of membrane potentials and opposes excitation, which actively delays firing of the next action potential, and thereby determines the duration of the interspike interval. The trajectory of this slow depolarization depends on the kinetics of ISA gating (Connor, 1978). More recent studies have corroborated this function of ISA in the nervous system of mammals and other organisms (Baxter and Byrne, 1991, Liss et al., 2001, Malin and Nerbonne, 2000, Shibata et al., 2000, Song et al., 1998, Tierney and Harris-Warrick, 1992). By controlling the interspike interval, ISA helps to determine the frequency-modulated code that is characteristic of neurons (Hille, 2001).

This review is mainly concerned with the physiological, genetic and molecular properties and the mechanisms of inactivation of the voltage-gated K+ channels that underlie ISA in the nervous system. We start by briefly describing the physiological properties of ISA and reviewing the family of genes that encode the ISA pore-forming subunit (α-subunit) and its specific structural properties. We then discuss the biophysical properties of the ISA pore forming subunit with emphasis on recent studies that have cast light on the molecular basis of ISA inactivation gating. Then, we increase the level of complexity, by reviewing the modulatory properties of novel accessory subunits (β-subunits) that are likely to be integral components of the ISA macromolecular complex. We conclude by reviewing other recent studies that have investigated the modulation of ISA by protein phosphorylation. Earlier papers published in the last two decades have reviewed the general properties of A-type K+ channels in neurons and myocytes from heart and smooth muscle (Nerbonne, 2000, Oudit et al., 2001, Rasmusson et al., 1998, Rogawski, 1985, Rudy, 1988).

In the whole-cell or dendritic cell-attached configurations, ISA is detectable at membrane potentials that are generally below the action potential threshold (Table 1). The voltage dependence of the peak K+ conductance has usually been described by assuming a Boltzmann function with the voltage of half-activation (V1/2) ranging between −47 mV (rat cerebellar granule cells) and 0 mV (hippocampal and neocortical pyramidal cells); and the slope factor (ka) that can range from 5 mV in rat cerebellar Purkinje cells to 27 mV in hippocampal inhibitory interneurons (Table 1). Upon a step depolarization, the outward ISA rapidly rises (τa ⋍ 1–5 ms), peaks, and usually decays monoexponentially (τ ⋍ 7–95 ms). In some instances, ISA inactivation exhibits a slow component (τ ⋍ 60–300 ms), which accounts for approximately 20–50% of the total current decay (Table 2). While the time constants of ISA inactivation generally do not exhibit significant voltage dependence, some studies have reported significant increases with voltage (Hoffman et al., 1997, Klee et al., 1995, Martina et al., 1998, Song et al., 1998).

An important feature of ISA is the hyperpolarized steady-state inactivation with a midpoint voltage (V1/2) ranging between −56 and −94 mV and a slope factor (ki) ranging between 6 and 10 mV, and therefore ISA is mostly inactivated at typical neuronal resting potentials (Table 2). Such a range of variability has also been observed in Drosophila and mammalian ISA (Baker and Salkoff, 1990, Tkatch et al., 2000, Tsunoda and Salkoff, 1995), which may reflect tissue differences in the subunit composition of the A-type K+ channel, association of modulatory subunits, posttranslational modifications, or environmental variations. Nevertheless, steady-state inactivation of ISA generally occurs at hyperpolarized membrane potentials (before appreciable activation), suggesting that a pathway of inactivation without channel opening (closed-state inactivation) plays an important role in ISA function, as reviewed later. Another distinguishing feature of ISA is its characteristic rapid recovery from inactivation at hyperpolarized membrane potentials, an essential feature of ISA. The majority of ISA recovers from inactivation with a time constant ranging between 10 and 30 ms at −110 mV. At more positive membrane potentials (−70 to −80 mV), recovery from inactivation may be biphasic with a prominent fast phase (τr = 10–40 ms) and a slower phase (τr = 329–962 ms) (Table 2).

Reviewing the pharmacological properties of ISA in detail is beyond the scope of this review. Here, we will briefly summarize some especially relevant aspects. Typically, ISA is inhibited by submillimolar and low millimolar concentrations of 4-aminopyridine (4-AP) and is relatively resistant to >10 mM tetraethylammonium (TEA). This pharmacological selectivity is broadly accepted and has been exploited in many studies to isolate the ISA component from the ensemble of K+ currents expressed in neurons (Bardoni and Belluzzi, 1993, Cull-Candy et al., 1989, Hoffman et al., 1997, Korngreen and Sakmann, 2000, Mitterdorfer and Bean, 2002, Schoppa and Westbrook, 1999, Solc and Aldrich, 1988, Song et al., 1998, Thompson, 1977, Tkatch et al., 2000). Cadmium (Cd++), a potent blocker of Ca++ channels, depresses ISA by positively shifting the voltage dependence of activation and inactivation gating (Mayer and Sugiyama, 1988). Exposure to 100 μM Cd++ can cause +10 mV shift in activation and +20 mV shift in inactivation, whereas exposure to 300 μM Cd++ shifts activation by +19 mV and shifts the apparent activation “threshold” by +10 mV (Song et al., 1998).

What is the molecular identity of somatodendritic A-type K+ channels? Molecular genetic approaches that began in Drosophila and expanded to various invertebrate and vertebrate species have established four gene subfamilies at the foundation of the superfamily of voltage-gated K+ channels (Kv channels). The Drosophila Shaker, Shab, Shaw, and Shal genes encode independent K+ current systems that are conserved throughout the animal kingdom (Butler et al., 1989, Covarrubias et al., 1991, Jan and Jan, 1997, Jegla and Salkoff, 1994, Salkoff et al., 1992, Wei et al., 1990, Wei et al., 1996). Presently, Kv1 (Shaker), Kv2 (Shab), Kv3 (Shaw), and Kv4 (Shal) are the accepted names of the corresponding subfamilies in vertebrate organisms (Gutman et al., 2003). In heterologous systems, single genes from Drosophila (Shaker and Shal) express rapidly inactivating K+ currents (Covarrubias et al., 1991, Wei et al., 1990), and several genes from the vertebrate Shaker, Shaw and Shal subfamilies (Kv1.4, Kv3.3, Kv3.4, Kv4.1, Kv4.2, and Kv4.3) also express rapidly inactivating K+ currents (Baldwin et al., 1991, Fernandez et al., 2003, Pak et al., 1991, Serodio et al., 1996, Stuhmer et al., 1989, Vega-Saenz de Miera et al., 1992). Additionally, accessory β-subunits of Kv1 channels (Kvβ1 and Kvβ3) can introduce fast inactivation in normally noninactivating, delayed-rectifier Kv1 channels (e.g., Kv1.1, Kv1.2, and Kv1.5) (Heinemann et al., 1995, Rettig et al., 1994).

Increasing amount of molecular, biochemical, immunological, pharmacological, and biophysical evidence now points to Kv4-type K+ channels underlying the somatodendritic A-type K+ current. First, all Kv4 protein products (Kv4.1, Kv4.2, and Kv4.3) are found in the mammalian nervous system, and Kv4 mRNA message is related to level of ISA in mammalian brain neurons. While in rat brain Kv4.2 and Kv4.3 are prevalent, all Kv4 proteins are clearly present in human brain (Isbrandt et al., 2000, Serodio and Rudy, 1998). In neostriatal medium spiny neurons and cholinergic interneurons, in globus pallidus neurons, and in basal forebrain cholinergic neurons, quantitative single-cell reverse transcription (RT) PCR and voltage clamping show that Kv4.2 mRNA abundance correlate linearly to ISA (Tkatch et al., 2000). In cultured cerebellar granule cells, ISA increases along with Kv4 protein expression over time (Shibata et al., 2000). Only Kv4.3L (long version), but not Kv4.1, Kv4.2, Kv1.4, or Kv3.4, is expressed in single dopaminergic substantia nigra neurons, and its expression correlates with the level of ISA (Liss et al., 2001). Second, in neurons, ISA can be suppressed by gene suppression technologies designed for Kv4 channels. Expression of Kv4 dominant-negative mutants in acute dissociated cultured cells eliminates ISA in cerebellar granule cells (Johns et al., 1997, Shibata et al., 2000) and ISA-fast in sympathetic neurons (Malin and Nerbonne, 2000).

Third, immunohistochemical analysis shows that Kv4.2 has a somatodendritic distribution (Sheng et al., 1992), and in adult hippocampus, Kv4.2 is expressed on distal dendrites and neuropils of CA1-3 neurons (Maletic-Savatic et al., 1995). The somatodendritic membrane of rat neostriatal cholinergic interneurons express Kv4.2 but not Kv1.4 according to immunocytochemical analysis (Song et al., 1998). Clustering of Kv4.2 channels can be found in the postsynaptic membrane (Alonso and Widmer, 1997). In contrast, Kv1.4/Kv1.2 channels are localized in the axons and nerve terminals, where they may be involved in the regulation of neurotransmitter release. In fact, the molecular mechanism behind specific targeting of Shal/Kv4 channels to dendrites is now understood. That is, a 16-amino-acid, dileucine-containing motif in Shal channels is responsible for dendritic targeting (Rivera et al., 2003) (Fig. 1). This motif is highly conserved from nematodes to humans, and when transplanted into Kv1.3 or Kv1.4, it alone is sufficient to target these axonal K+ channels to dendrites.

Fourth, Kv4 currents and ISA have very similar pharmacological profiles (Bardoni and Belluzzi, 1993, Cull-Candy et al., 1989, Hoffman et al., 1997, Korngreen and Sakmann, 2000, Mitterdorfer and Bean, 2002, Schoppa and Westbrook, 1999, Solc and Aldrich, 1988, Song et al., 1998, Thompson, 1977, Tkatch et al., 2000). Both Kv4 current and ISA exhibit insensitivity to external TEA up to 20 mM and suppression by 4-AP with a half-inhibition concentration at the low millimolar levels, and Cd++ induces depolarizing shifts in the voltage dependence of steady-state activation and inactivation. Other studies reported that both Kv4.2 and ISA are inhibited by arachidonic acid (Dryer et al., 1998, Keros and McBain, 1997, Villarroel and Schwarz, 1996). Finally, as discussed through the rest of this review, the biophysical and electrophysiological properties of Kv4 channels and the ways in which specific auxiliary subunits remodel them reinforces the idea of a direct relationship between Kv4 channels and the somatodendritic A-type K+ channels that underlie ISA.

Kv4 potassium channels, like voltage-gated K+ channels, are formed from tetrameric complexes of identical or genetically related α-subunits from the same subfamily. All Kv α-subunits possess a cytoplasmic amino-terminal region, six transmembrane segments (S1–S6) plus their associated interconnecting intracellular and extracellular loops, and a cytoplasmic carboxy-terminal region. The subfamily-specific homo- or heterotetramerization of α-subunits occurs despite the close evolutionary relationship between Kv1 (Shaker), Kv2 (Shab), Kv3 (Shaw), and Kv4 (Shal) genes (Covarrubias et al., 1991), which share approximately 40% identity within the highly conserved S1–S6 region (Salkoff et al., 1992). It is not surprising that the S1–S6 region, known as the “core”, conducts the main businesses of potassium selectivity, ion conduction, and voltage-dependent gating. The S5 and S6 segments form the innermost structures and line the pore region through which K+ must traverse down its electrochemical gradient across the membrane. The S5–S6 loop, or pore (P)-loop, sits at the outer end of the “inverted teepee” formed by S5 and S6 segments from all four α-subunits and acts as the selectivity filter (Doyle et al., 1998). Carbonyl oxygen atoms from the K+ channel signature sequence (TTXGYGD) line the selectivity filter, where they preferentially coordinate K+. At the other end of this inverted teepee, the constricted region created by the convergence of inner S6 segments (helix bundle crossing, corresponding to the “smokehole” of an “inverted teepee”) provides the ideal site for controlling gating of the K+ channel. The S6 segment is highly conserved within the animal branch of the Kv family, and a pair of highly conserved prolines (PxP motif) is likely to bend the helical S6 in both the open and closed conformations (del Camino and Yellen, 2001, del Camino et al., 2000). X-ray crystallographic studies using tetrameric, two-transmembrane domain channels from bacteria (KcsA and MthK) suggest how S6 movements can physically close and open the channel (Jiang et al., 2002). The bacterial M2 segments, homologues of the Kv S6 segments, can open the channel by swinging around a critical glycine hinge and separating the helix bundle. It is possible that combined displacements involving the glycine hinge and the conserved PxP motif underlie gating in Kv channels (Webster et al., 2004). The voltage-sensing portions of Kv channels concern the S1–S4 segments, which probably surround the pore. The S4 segment is often called the “voltage sensor”, which includes several positively charged arginines at every third position. The first 3–4 arginines at the N-terminal side of S4 provide the charges that move across the electric field during voltage-dependent gating (approximately 12–16 charges per tetramer) (Bezanilla, 2002, Horn, 2000, Sigworth, 1994).

In contrast to the highly conserved core region, the cytoplasmic N- and C-terminal portions diverge markedly among Kv channels. Within a subfamily, a highly conserved T1 domain located in the hydrophilic amino-terminal region immediately preceeding the S1 segment is the minimal requirement for subfamily-specific tetramerization of α-subunits (Li et al., 1992, Shen et al., 1993, Xu et al., 1995). Determination of the crystal structure of Aplysia Shaker T1 reveals four T1 domains arranged in fourfold symmetry surrounding a central opening (Kreusch et al., 1998). Polar interactions at the tetramerization interfaces provide the basis of subfamily-specific assembly (Bixby et al., 1999, Kreusch et al., 1998, Minor et al., 2000). Additionally, while the crystal structures of Kv1, Kv3, and Kv4 T1 tetramers all display a common four-layered scaffolding, the Kv3 and Kv4 T1 tetramers contain four Zn++ atoms that are coordinated between adjacent subunits at the tetramerization interfaces within layer 4 (Bixby et al., 1999, Nanao et al., 2003, Scannevin et al., 2004). Structural studies localize the T1 tetramer directly below the inner vestibule, where it figuratively exists as a “hanging gondola” (Gulbis et al., 2000, Kobertz et al., 2000, Orlova et al., 2003, Sokolova et al., 2001). K+ must pass through four side “windows” possibly framed in part by the T1-S1 linkers and the intracellular C-terminal regions.

In mammals, three different Kv4 genes have been discovered, Kv4.1, Kv4.2, and Kv4.3 (Baldwin et al., 1991, Pak et al., 1991, Serodio et al., 1996), and Kv4.3 is the only member of this family that exhibits alternative splicing in mammals (Kv4.3-long and Kv4.3-short) (Dilks et al., 1999) (Fig. 1). Structurally and functionally, Shal K+ channels are highly conserved across the animal kingdom (Isbrandt et al., 2000, Salkoff et al., 1992) (Fig. 2). The amino acid identity for the core region (S1–S6) is 82% between fly and rodent Shal/Kv4 channels, as compared with 62–71% for Shaker/Kv1, 75% for Shab/Kv2, and 51–52% for Shaw/Kv3 (Pak et al., 1991, Salkoff et al., 1992). Amino acid comparison reveals that Kv4.3 is more similar to Kv4.2 (73% identity) and Drosophila Shal (71% identity) than with Kv4.1 (63% identity). A complete evolutionary tree of Shal channels and an alignment of representative members are presented in Fig. 1, Fig. 2.

Compared to other members of Shaker superfamily of Kv channels, Shal K+ channels exhibit a high degree of structural and functional conservation among all known members found across the animal kingdom, from the jelly fish and nematodes to humans (Salkoff et al., 1992). Although Shal K+ channels share the key structural motifs that are characteristic of all Kv channels (e.g., the positively charged S4 voltage sensor, the TTXGYGD signature sequence in the selectivity filter, and the PxP motif in the S6 segment), they have specific and highly conserved structural features that define their function. First, the proximal N-terminal region (positions 2–40) is mostly hydrophobic and exhibits a high degree of amino acid identity among all known Shal channels. This region constitutes an important binding determinant of auxiliary subunits known as KChIP-x (Bähring et al., 2001b, Scannevin et al., 2004, Zhou et al., 2004). In contrast to the equivalent regions in K+ channels that undergo classical N-type inactivation (Drosophila ShakerB, Kv1.4, and Kv3.4), only a single positively charged amino acid resides within the first 20 amino acids of Shal/Kv4 subunits (discussed in more detail below). The N-terminal segment that encompasses residues 45–140 constitutes the specific tetramerization domain (also known as T1; see general Kv properties above) that limits the coassambly of pore-forming subunits to members of the Shal subfamily (Covarrubias et al., 1991, Li et al., 1992, Nanao et al., 2003, Pfaffinger and DeRubeis, 1995, Strang et al., 2001). An interesting feature of this region is the presence of a Zn++ binding site that is also present in Kv2 and Kv3 channels but absent in Kv1 channels (Bixby et al., 1999, Nanao et al., 2003). Zn++ binding occurs between subunits and appears to be critical for the assembly and conformational stability of Kv4 channels (Jahng et al., 2002, Strang et al., 2003). In light of possible contributions of the T1 interfaces to Kv channel gating (Cushman et al., 2000, Minor et al., 2000), it is conceivable that the Zn++ binding site in Kv4 channels may also have a specific functional contribution. Another outstanding feature of Shal channels is found in the S4–S5 loop, a cytoplasmic linker thought to be a critical element of the activation gate in Kv channels. Here, the unique motif KSCASE is absolutely conserved in all Shal channels. Mutations in this region drastically affect deactivation and inactivation in unexpected ways (Jerng et al., 1999). Shal-specific motifs are also found in regions that surround the selectivity filter. In contrast to most prokaryotic and eukaryotic K+ channels that share the WW motif at the N-terminal side of the S5–S6 P-loop (Doyle et al., 1998, Jiang et al., 2003, Kuo et al., 2003), all Shal K+ channels share the WY sequence at the corresponding positions. From the crystal structure of KcsA and other bacterial channels (Doyle et al., 1998, Jiang et al., 2003, Kuo et al., 2003), it is clear that the WW motif is part of an aromatic cuff that stabilizes the selectivity filter and helps to determine the pore size (Doyle et al., 1998). We hypothesize that the distinct Shal-specific WY motif along with other specific P-loop differences may have allowed alternative mechanisms of inactivation to evolve. Additionally, valine and serine occupy high impact P-loop and S6 positions in Shal K+ channels, where they would strongly destabilize P/C-type inactivation at equivalent positions in Shaker K+ channels (Hoshi et al., 1991, Lopez-Barneo et al., 1993) (Table 5). The post-S6 cytoplasmic region (approximately 30 residues) of Shal K+ channels is also conspicuously rich in basic amino acids (e.g., the highly conserved segment RADKRXAQRKARLAR, where X is R or K), which may contribute to activation and specialized inactivation gating (Hatano et al., 2003, Jerng and Covarrubias, 1997). Lastly, further downstream in the cytoplasmic C-terminal region, there is a Shal-specific HHHLL motif that, among the known invertebrate and vertebrate Shal K+ channels, is only absent in the jellyfish Shal orthologue. The LL motif is critical for the somatodendritic localization of Shal K+ channels in the nervous system, but the function of the HHH sequence is not yet known. Among all known Shal K+ channels, the outstanding structural identity of the specialized regions summarized above clearly suggests that the corresponding A-type K+ channels share the same mechanisms that confer integrative and signal coding properties of neurons of most animals.

Many K+ channels autoregulate their function by a process known as inactivation. In heterologous expression systems, Kv channels undergo fast inactivation that develops over tens of milliseconds, and slower inactivation that may take hundreds of milliseconds to tens of seconds to develop. Fast inactivation is found in all known Shal/Kv4 channels (Baldwin et al., 1991, Baro et al., 1996, Jegla and Salkoff, 1997, Pak et al., 1991, Santi et al., 2003, Serodio et al., 1996, Zhu et al., 1999) but only in certain mammalian members of the Shaker/Kv1 and Shaw/Kv3 families (Schroter et al., 1991, Stuhmer et al., 1989, Vega-Saenz de Miera et al., 1992); and some Kv1 channels may adopt fast inactivation from auxiliary β subunits (Heinemann et al., 1994, Heinemann et al., 1995, Rettig et al., 1994). Thus, it appears that mammalian Kv1 and Kv3 channels evolved to conduct a wide variety of functions (with only some of them requiring fast inactivation) and Kv4 channels underwent specialization in early invertebrates to conduct a fundamental function in the nervous system that was probably required for the integration of sensory inputs (Salkoff et al., 1992).

Inactivation of voltage-gated ion channels may originate from open or closed states (Aldrich et al., 1983, Ayer and Sigworth, 1997, Bean, 1981, Greenstein et al., 2000, Patil et al., 1998, Roux et al., 1998, Solc and Aldrich, 1990, Zagotta and Aldrich, 1990) (Scheme 1). Generally, ShakerB, Kv1.3, Kv1.4, Kv3.3, and Kv3.4 channels exhibit preferential open-state inactivation (Beck et al., 1998, Fernandez et al., 2003, Hoshi et al., 1990, Panyi et al., 1995, Yellen, 1998, Zagotta and Aldrich, 1990), which implies that at depolarized membrane potentials inactivation is significantly coupled to channel opening (i.e., the channel must open before it can inactivate). In contrast, Kv2 and Kv4 channels exhibit preferential closed-state inactivation (Bähring et al., 2001a, Beck and Covarrubias, 2001, Jerng et al., 1999, Klemic et al., 1998). In this case, a channel may begin to inactivate before it opens, causing cumulative inactivation upon repetitive stimulation (Aldrich et al., 1979, Klemic et al., 1998). Furthermore, if channel opening is a difficult transition (i.e., the opening step is backward biased or weakly forward biased) when the voltage sensors in the channel are fully activated, then inactivation may preferentially occur from the preopen closed state (Fig. 3A, scheme). Preferential closed-state inactivation can explain the macroscopic inactivation properties of all Kv4 channels (Bähring et al., 2001a, Beck and Covarrubias, 2001, Beck et al., 2002, Gebauer et al., 2004, Jerng et al., 1999, Shahidullah and Covarrubias, 2003). For instance, relatively rapid recovery from steady-state inactivation without channel reopening suggested that upon a prolonged depolarization, Kv4 channels accumulate in a deep inactivated state that is directly connected to a closed state (Bähring et al., 2001a, Gebauer et al., 2004). Contrary to a prediction from open-state inactivation, the observed macroscopic inactivation of Kv4.1, Kv4.2, and Kv4.3 can be indirectly delayed by Rb+ through its “foot-in-the-door” effect on channel closing (only when Rb+ carries the current) (Bähring et al., 2001a, Shahidullah and Covarrubias, 2003, Swenson and Armstrong, 1981). ISA also exhibits evidence of preferential closed-state inactivation. In several instances, the observed time constant of macroscopic ISA inactivation increases with membrane depolarization over the range of voltages where the current activates (Hoffman et al., 1997, Klee et al., 1995, Martina et al., 1998, Song et al., 1998). This apparent inverse relationship between inactivation rate and voltage can be explained by a declining probability of residence in the preopen inactivation-permissive closed state(s) caused by a gradual increase in open probability at more positive membrane potentials.

All Kv4 α subunits expressed in heterologous expression systems (Xenopus oocytes or mammalian cells) induce outward K+ currents that peak quickly and inactivate following a complex time course, which is well described by assuming the sum of two to three exponential terms (Table 3). The time constants of inactivation exhibit little or no voltage dependence and, upon a prolonged depolarization (1–5 s), Kv4 inactivation is nearly complete (<Fsteady-state; Table 3). This behavior appears to be an intrinsic property of Kv4 α-subunits because a similar macroscopic complexity is observed when the Kv4 currents are examined using various voltage-clamping methods (Bähring et al., 2001a, Beck and Covarrubias, 2001, Beck et al., 2002, Jerng and Covarrubias, 1997, Jerng et al., 1999, Pak et al., 1991, Shahidullah and Covarrubias, 2003, Wang et al., 2002). The main difference between the inactivation kinetics of Kv4 channels is the relative weight of the exponential terms. While inactivation of Kv4.2 and Kv4.3 currents is dominated by the fast phase, inactivation of Kv4.1 currents is dominated by the slower phases. From kinetic analysis alone, it is generally not possible to unambiguously establish a quantitative correspondence between the phases of macroscopic inactivation and specific pathways of inactivation. However, mutational and kinetic analyses have generally associated the fast phase to open-state inactivation and the intermediate or slower phase to inactivation from a preopen closed state (Bähring et al., 2001a, Beck and Covarrubias, 2001, Jerng and Covarrubias, 1997, Jerng et al., 1999). Interestingly, upon patch excision in Xenopus oocytes, the loss of a soluble cytoplasmic factor accelerated inactivation of Kv4.1 and Kv4.3 channels by reducing the time constants of the decay and enhancing the relative magnitude of the fast phase (Beck and Covarrubias, 2001). Such changes along with a reduced peak current was economically explained by a selective acceleration of preferential closed-state inactivation (Beck and Covarrubias, 2001).

The kinetic complexity of macroscopic Kv4 inactivation at positive membrane potentials in heterologous expression systems appears to be greater than that of native ISA (Table 2, Table 3). In spite of this apparent discrepancy, all Kv4 currents undergo deep steady-state closed-state inactivation at subthreshold membrane potentials. Under relatively intact conditions (whole-cell or cell-attached patch clamp and two-electrode voltage clamp), the V1/2 of prepulse steady-state inactivation of Kv4 currents ranges between −56 and −70 mV, which agrees with values derived from native ISA (Table 2, Table 3). In 16/18 different studies of ISA, V1/2 ranges between −60 and −93 mV. In inside-out patches, the V1/2 of Kv4 currents shifts approximately 10 mV in the hyperpolarized direction, which is the expected change when closed-state inactivation is accelerated (as described above; Beck and Covarrubias, 2001). One of the major discrepancies between native and expressed Kv4 currents resides, however, in the rate of recovery from inactivation. The time constant of recovery from inactivation of Kv4 currents at hyperpolarized membrane potentials ranges between 100 and 200 ms (Table 3), which is significantly slower than that of native ISA (<60 ms for the dominant kinetic component of the removal of inactivation; Table 2).

Kv4 α-subunits expressed in heterologous expression systems also generate currents with similar activation properties. Kv4 outward currents are first detectable at membrane potentials between −50 and −40 mV and a typically asymmetric peak conductance–voltage relation is generally better described by assuming a fourth-order Boltzmann function (Bähring et al., 2001a, Beck and Covarrubias, 2001, Beck et al., 2002, Jerng and Covarrubias, 1997, Jerng et al., 1999, Pak et al., 1991, Shahidullah and Covarrubias, 2003). The midpoint voltage of the peak conductance–voltage relation (VG=50%) ranges between −3 and −10 mV, which is within the range of values observed for native ISA in at least five different neuronal types (Table 1). In most neurons though, this parameter is more hyperpolarized (Table 1), suggesting the influence of auxiliary subunits and other modulatory factors, as reviewed later. Under physiological conditions (2 mM external K+), tail current relaxations of Kv4 currents at hyperpolarized voltages (deactivation) are very fast. The dominant time constant of Kv4 deactivation is of the order of 1–5 ms (at −120 mV; Table 3). Although there is only limited information on the deactivation of native ISA, some representative studies suggest that it is complete in less than 5 ms (Bardoni and Belluzzi, 1993, Belluzzi et al., 1985, Cull-Candy et al., 1989, Korngreen and Sakmann, 2000). Additionally, there is limited information on the single channel properties of ISA, but several studies have determined the unitary conductance of channels generated from recombinant Kv4 α-subunits (Beck et al., 2002, Holmqvist et al., 2002, Jerng et al., 1999). These studies reported complex fast gating and conspicuous fluctuations of single channel currents between 2 and 3 distinct conductance levels. The main Kv4 unitary conductance ranges between 4 and 5 pS and is similar to that of Drosophila Shal channels (Tsunoda and Salkoff, 1995), but is lower than that determined from native ISA in two other studies (approximately 7 pS) (Hoffman et al., 1997, Solc and Aldrich, 1988). Frequent subconductance levels ranged between 1.5 and 3 pS (Beck et al., 2002). Measurements of gating currents of Kv4 channels are not yet available.

For more than a decade, many studies have investigated two distinct molecular mechanisms of inactivation in voltage-gated K+ channels (Aldrich et al., 1990, Choi et al., 1991, Demo and Yellen, 1991, Gulbis et al., 2000, Holmgren et al., 1996, Hoshi et al., 1990, Hoshi et al., 1991, Liu et al., 1996, Rasmusson et al., 1998, Yellen, 1998, Yellen, 2002, Zhou et al., 2001). These studies mainly concerned the Drosophila Shaker K+ channel, its closely related mammalian Kv1 homologues and the Kv1-β1 subunit. It is thus broadly accepted that ShakerB/Kv1 channels undergo N-type and P/C-type inactivation, which typically correspond to fast and slow inactivation processes, respectively. Elegantly, electrophysiological, mutational, and structural approaches have converged to yield a detailed mechanistic picture of N-type inactivation, which is determined by an amphiphilic segment that constitutes the first 20–30 amino acids of the pore-forming subunit (the inactivation domain: ID) (Aldrich, 2001, Gulbis et al., 2000, Zhou et al., 2001). Basically, long-range electrostatic interactions drive the positively charged ID (four per tetramer) to the inner mouth of the channel (Murrell-Lagnado and Aldrich, 1993a, Murrell-Lagnado and Aldrich, 1993b). Acidic amino acids that contribute to this interaction are located in the T1-S1 segment, within a region that helps to form lateral access windows at the internal face of the channel (where the transmembrane core joins the cytoplasmic T1 domain) (Gulbis et al., 2000). Once at the inner mouth of the channel, the nonpolar moiety of a single ID adopts an extended chain conformation that “snakes” into the pore and occludes it upon forming relatively stable hydrophobic interactions with residues that line the pore (MacKinnon et al., 1993, Zhou et al., 2001). Conclusive evidence of classical N-type inactivation is also available for Kv1.4, Kv3.3, and Kv3.4 (Antz et al., 1997, Beck et al., 1998, Covarrubias et al., 1994, Fernandez et al., 2003, Lee et al., 1996, Murrell-Lagnado and Aldrich, 1993b, Ruppersberg et al., 1991, Schroter et al., 1991, Tseng-Crank et al., 1993); and for auxiliary subunits Kvβ1 (Kv1-specific) and the β3 subunit of the large-conductance Ca++-activated K+ channel (Bentrop et al., 2001, Gulbis et al., 2000, Wissmann et al., 1999).

The presence of classical N-type inactivation in Kv4 channels has remained uncertain (Table 4). In Kv4.1, the deletion Δ2–32 or Δ2–71 mainly eliminated a relatively fast (τ = 18 ms) but small component (approximately 20%) of a complex time course of inactivation (Jerng and Covarrubias, 1997). Similarly, in Kv4.2, the deletion Δ2–40 slowed the fast phase of inactivation and also decreased its relative weight (τ(wild-type) = 11 ms and τ(Δ2–40) = 35 ms, contributing 73% and 33%, respectively) (Bähring et al., 2001a). In both instances, however, inactivation of the deletion mutants was nearly complete at the end of a 1-s pulse to +50 mV, as observed with the wild-type counterparts. Although these initial observations appeared consistent with N-type inactivation, other results did not agree with the classic features of this mechanism as observed in ShakerB K+ channels or heteromeric channels made of Kv1 channels and Kvβ1 (Table 4). For instance, inactivation of Kv4.1 channels is not affected by 5 mM internal TEA, although there is a substantial inhibition of the peak current (approximately 50%) (Jerng and Covarrubias, 1997). Within the sequence of the first 20 residues (mostly hydrophobic) there is only one highly conserved basic amino acid at position R13. Contrary to the predictions of classical N-type inactivation, however, charge neutralization (R13Q) or reversal (R13E) at this position had no effect on the development of Kv4.1 and Kv4.2 inactivation (Jerng and Covarrubias, 1997, Gebauer et al., 2004). Additionally, the mutations K34Q+R37Q in Kv4.1 and R35Q+K36Q in Kv4.2 had no impact on the development of inactivation (Baldwin et al., 1991, Jerng and Covarrubias, 1997). However, the deletion Δ35–37 significantly slowed Kv4.2 inactivation (Baldwin et al., 1991). Thus, long-range electrostatic interactions are not necessary for fast Kv4 inactivation but the length of the N-terminus is clearly important. A study that investigated the docking sites of the ShakerB ID found that mutations of a highly conserved glutamate in the S4–S5 loop (E392Q or E392D) eliminated inactivation (Isacoff et al., 1991). In contrast, equivalent mutations in Kv4.1 (E325Q or E325D) dramatically accelerated inactivation (Jerng and Covarrubias, unpublished). Two crucial observations demonstrated that N-type inactivation in ShakerB, Kv1.4, and Kv3.4 channels results from an open-pore occlusion by the ID. First, these channels reopen during the recovery from N-type inactivation; and second, incoming K+ accelerates the recovery from inactivation by directly displacing the ID out of the pore (the “knock-off” effect) (Demo and Yellen, 1991, Ruppersberg et al., 1991). Two independent attempts to test N-type inactivation of Kv4.1 and Kv4.2 in the presence of elevated external K+ failed to demonstrate an open-pore blocking mechanism mediating inactivation (Bähring et al., 2001a, Jerng and Covarrubias, 1997). An exceptional case is jShal-γ1, a jelly fish regulatory subunit with a positively charged N-terminal ID domain (+7, from 1 to 30) and other specific features in S4–S5 and the S6 tail (possible docking sites of the ID) that confer fast N-type inactivation in heterologously expressed Shal/Kv4 channels (Jegla and Salkoff, 1997). However, jShal-γ1 does not produce functional channels by itself and no vertebrate homologues of jShal-γ1 have been found.

Surprisingly, certain C-terminal deletions in Kv4.1 induced a current phenotype that closely mimics the elimination of fast inactivation by the N-terminal deletion Δ2–71, which suggested that the cytoplasmic N- and C-termini may act in concert to determine fast inactivation (Jerng and Covarrubias, 1997). In contrast, C-terminal deletions do not significantly affect inactivation of ShakerB K+ channels (Hoshi et al., 1991). In spite of all the discrepancies discussed above, it is conceivable that Kv4 channels share features of N-type inactivation that are confounded by a dominant preferential closed-state inactivation. In fact, a peptide corresponding to the first 20 amino acids of Kv4.2 can confer fast inactivation when applied to the intracellular side of patches expressing the Kv4.2 (Δ 2–40) (Gebauer et al., 2004). However, compared to the ShakerB ID, the putative Kv4 N-terminal ID may have a much lower affinity for the internal lining of the pore (Gebauer et al., 2004). Such a weak interaction may be responsible for the relatively unstable open-state inactivation assumed for all Kv4 channels (Bähring et al., 2001a, Beck and Covarrubias, 2001, Beck et al., 2002, Jerng et al., 1999). Unstable open-state inactivation may be especially significant in Kv4.1 channels, which exhibit a relatively small fast phase of inactivation (≤20%; in cell-attached and whole-oocyte configurations) (Beck and Covarrubias, 2001, Beck et al., 2002, Jerng and Covarrubias, 1997). A notable specific feature of Kv4.1 (Fig. 1) is the presence of threonine-7 and valine-17 instead of more hydrophobic alanine and isoleucine at the respective sites of more rapidly inactivating Shal/Kv4 channels (e.g., nShal, lShal, dShal, Kv4.2, and Kv4.3). The pore determinants that underlie fast unstable inactivation of Kv4 channels by a low-affinity occlusion at the inner mouth of the pore need to be elucidated.

In contrast to N-type inactivation, the molecular mechanism of P/C-type inactivation in Kv channels is less well understood. The observed rates of C-type inactivation in these channels can differ by 2–3 orders of magnitude (50–0.05 s−1). Most studies, however, agree that P/C-type inactivation probably involves a concerted conformational change that constricts the outer mouth of the pore and is likely to involve the selectivity filter of the channel (Liu et al., 1996, Ogielska et al., 1995, Panyi et al., 1995, Yellen, 1998). The speed of this conformational change can be influenced by the occupancy of the pore by permeant ions or by an external pore blocker. Thus, when the concentration of external K+ is increased or the channels are exposed to external TEA, P/C-type inactivation is slowed by a “foot-in-the-door” mechanism (Choi et al., 1991, Lopez-Barneo et al., 1993). Conversely, when the pore becomes vacant as a result of an internal pore occlusion by N-type inactivation (or an internal blocker), P/C-type inactivation is drastically accelerated (Baukrowitz and Yellen, 1995, Baukrowitz and Yellen, 1996). P/C-type inactivation is partially coupled to N-type inactivation, and therefore becomes slower when N-type inactivation is removed (Hoshi et al., 1991). It is hypothesized that faster P/C-type inactivation in a vacant pore may be caused by a constriction of the selectivity filter (Yellen, 1998, Yellen, 2002). In agreement with these observations, P/C-type inactivated channels are relatively more permeable for Na+, which suggests a reduced pore diameter (Doyle et al., 1998, Kiss et al., 1999, Starkus et al., 1997). Some studies have referred to P-type inactivation to describe the apparent pore constriction that underlies inactivation (De Biasi et al., 1993, Loots and Isacoff, 1998), and hypothesized that the activated voltage sensors (S4 segment) undergo an additional rearrangement that helps to stabilize C-type inactivation through a direct interaction with the pore (Loots and Isacoff, 1998, Olcese et al., 1997).

The presence of classical P/C-type inactivation in Kv4 channels is also doubtful (Table 5). Critically, and contrary to the predictions of a “foot-in-the-door” mechanism that impedes C-type inactivation, elevated external K+ accelerated hippocampal ISA, Kv4.1 and Kv4.3 inactivation (Jerng and Covarrubias, 1997, Kirichok et al., 1998, Shahidullah and Covarrubias, 2003). This observation could not be explained by the typically hyperpolarized voltage-dependent gating induced by elevated external K+ because the development of Kv4 inactivation over the voltage range of current activation exhibits little or no voltage dependence. Moreover, at voltages where the peak conductance is leveling off the development of inactivation remains faster in the presence of elevated external K+ (Shahidullah and Covarrubias, 2003). In agreement with the absence of P/C-type inactivation, 96 mM external TEA blocked the Kv4.1 current by approximately 30% but had no effect on the development of inactivation (Jerng and Covarrubias, 1997), and there are no apparent changes in relative ionic selectivity when Kv4.1 channels inactivate (Shahidullah and Covarrubias, 2003). Additionally, in contrast to the typically very slow recovery from C-type inactivation (many seconds at room temperature), recovery from inactivation is 1–2 orders of magnitude faster in Kv4 channels. This fast recovery from inactivation is slowed or unaffected by elevated external K+ (Jerng and Covarrubias, 1997) (contrary to the accelerated recovery from inactivation of Kv1 channels in elevated external K+) (Gomez-Lagunas and Armstrong, 1994, Levy and Deutsch, 1996). Slower recovery from inactivation and faster closed-state inactivation may account for a substantial hyperpolarized shift (ΔV1/2 ≈ −25 mV) induced by elevated external K+ on all Kv4 channels (Shahidullah and Covarrubias, 2003).

It is generally accepted that the determinants of P/C-type inactivation are located in regions that contribute to the pore and the selectivity filter of Kv channels. Thus, specific differences between the pore regions of Kv4 and other Kv subunits may reasonably explain the apparent absence of classical P/C-type inactivation in Shal/Kv4 channels. For instance, the mutation T449V in the ShakerB K+ channel practically eliminates C-type inactivation (Lopez-Barneo et al., 1993). The equivalent position in all Shal/Kv4 channels is occupied by valine (V374 in Kv4.3). Nevertheless, all Shal/Kv4 channels are capable of undergoing fast inactivation that develops in tens to hundreds of milliseconds. Very fast P/C-type inactivation has been observed in ShakerA K+ channels (τ ≈ 20 ms) because a key position in the proximal segment of S6 is occupied by valine (V463), a residue that indirectly decreases K+ affinity at a critical site in the pore and thereby accelerates C-type inactivation (Hoshi et al., 1991, Ogielska and Aldrich, 1998, Ogielska and Aldrich, 1999, Ogielska et al., 1995) (S388 at the equivalent position in Kv4.3 and equivalent sites in all Shal/Kv4 channels). However, a double ShakerB mutant (T449V / A463V) also exhibits very slow C-type inactivation because the effects of the T449V mutation are dominant (Ogielska et al., 1995). To explain the accelerating effect of elevated external K+ on Kv4 channel inactivation, it is interesting to speculate that the structural stabilization of the selectivity filter by elevated external K+ may favor a conformational change that underlies preferential closed-state inactivation at another location (see below). An apparent pore collapse in Kv4.3 channels exposed to low external K+ probably represents a general structural destabilization that is a hallmark of all vacant K+-selective pores (Eghbali et al., 2002, Ogielska et al., 1995). This epiphenomenon is probably distinct from the native mechanism of Kv4 inactivation that is observed under more physiological conditions.

A clue that guided the initial studies on the structural basis of Kv4 inactivation was the realization of a mutually exclusive interaction between 4-AP binding and inactivation of cardiac ITO (mediated by Kv4.2 or Kv4.3 subunits) and heterologously expressed Kv4.2 (Campbell et al., 1993, Tseng, 1999, Tseng et al., 1996). Mainly, these studies suggested that closed channels could either inactivate or bind 4-AP and that channel opening caused drug dissociation. Thus, it was supposed that mutations of internal residues that influence 4-AP binding may also impact inactivation and the stability of the open state (Jerng et al., 1999). Accordingly, mutations of two highly conserved valines to isoleucine (V404I and V406I) in the distal section of the S6 segment in Kv4.1 dramatically slowed inactivation and deactivation (channel closing) and reduced the inhibition by 4-AP. The novel phenotype of these mutants was reminiscent of that of the Kv2.1 channel, which has isoleucines at those positions and normally exhibits very slow inactivation, slow deactivation and low-affinity for 4-AP (Kirsch et al., 1993). Incidentally, the same study found that mutation of a highly conserved cysteine (C322S) in the S4–S5 loop of Kv4.1 also dramatically also slowed deactivation and inactivation. To explain how the mutations concomitantly slowed channel closing and inactivation, one can simply assume that the opening equilibrium of Kv4 channels is not strongly forward biased and that the preopen closed state is inactivation permissive (Fig. 3). Interestingly, analogous S6 mutations in the Kv1.4 channel, which undergoes preferential open-state inactivation, did not affect the development of inactivation but slowed channel closing and recovery from inactivation (Jerng et al., 1999). In contrast, equivalent mutations in Kv4.3 produced effects qualitatively similar to those observed in Kv4.1 (Wang et al., 2002). Therefore, preferential closed-state inactivation of Kv4 channels at all relevant membrane potentials economically explains the results of the mutational analysis reviewed above (Jerng et al., 1999), inactivation gating of Kv4.2 (Bähring et al., 2001a), modulation of closed state inactivation in Kv4.1 and Kv4.3 (Beck and Covarrubias, 2001, Beck et al., 2002, Wang et al., 2002), and the link between ion permeate on and inactivation gating in all Kv4 channels (Shahidullah and Covarrubias, 2003).

Although there is still much to be learned about the mechanisms of inactivation of Shal/Kv4 channels, the information gathered so far supports the following conclusions: (1) The N-terminal ID can confer rapid, but unstable, open-state inactivation in Kv4 channels; (2) elements that are critical for coupling the voltage sensor to the activation gate (S4–S5 loop and the distal segment of S6) are determinants of Shal/Kv4 inactivation observed in the absence of an N-terminal ID; (3) the opening equilibrium of Shal/Kv4 channels is not strongly forward biased; and (4) Shal/Kv4 channels undergo preferential closed-state inactivation. The latter two conclusions imply that at voltages that induce maximal activation a channel fluctuates between the open and preopen closed states and must close before it inactivates. As a working hypothesis, we propose that the Shal/Kv4-specific motif in the S4–S5 loop (KSCASE) may prevent effective coupling between the movements of the voltage sensor to the activation gate (the distal S6 segment). Consequently, channel opening is not a favorable transition. As proposed for the hyperpolarization-activated cation channel (HCN), when the sensors are activated but the channel is still closed, the S4–S5 loop may “slip” and the activation gate fails to open (Shin et al., 2004). Such a “slippage” of the S4–S5 loop may be the structural shift that underlies preferential closed-state inactivation of Shal/Kv4 channels. In light of these considerations, closed-state inactivation of Kv channels in general may be redefined as the failure of a channel to open or desensitization to voltage (Shin et al., 2004). Alternatively, a modified P/C-type mechanism of inactivation may exist in Kv4 channels. However, to explain the current data, such a mechanism needs to invoke distinct K+ affinities and repulsive ion–ion interactions in the Kv4 pore that negate the “foot-in-the-door” effect (see above) and may facilitate inactivation from the closed state (Eghbali et al., 2002).

The inherent properties of ion channels commonly undergo significant fine-tuning, or modulation, by regulatory factors or enzymatic processes. Binding of endogenous ligands, association with protein partners, or covalent modification by enzymes can modify the channel's structure to regulate its function. In general, modulation of ion channels can be observed at several levels, including the number of functional channels, the probability of opening, and the conductance of single channels.

A comparison of the properties of native and heterologously expressed Kv4 channels reveals remodeling and modulation at multiple levels. For instance, Kv4.2 channels expressed in cultured mammalian cells tend to collect intracellularly in the perinuclear ER, but acutely dissociated hippocampal neurons show robust localization of Kv4 throughout the soma and dendrites surfaces (Shibata et al., 2003). Second, although bearing qualitative resemblance to ISA, Kv4 currents in heterologous cells display functionally significant functional differences from their native ISA counterparts (Table 1, Table 2, Table 3). These differences, especially the faster kinetics of inactivation and recovery from inactivation, cannot be explained by heterotetramerization of Kv4 α-subunits alone. Third, the classical somatodendritic ISA currents from different neuronal cell types possess varying waveform kinetics and voltage-dependent activation and inactivation properties, suggesting that ISA itself may be undergoing cell-specific modulation (Table 1). Yet, despite variability in properties, the typical ISA consistently activates, inactivates, and recovers from inactivation faster than Kv4 currents heterologously expressed in mammalian cells or oocytes. Recently, significant advances have been made to understand the remodeling and modulation of Kv4 channel gating as reviewed below.

Auxiliary β-subunits play important modulatory roles on surface expression and gating of voltage-dependent K+, Na+, and Ca++ channels (Isom et al., 1994, Rettig et al., 1994). Ion channel β-subunits appear to come in two distinct flavors: cytosolic proteins that attach themselves to the cytoplasmic portion of the channel (e.g., Kvβ), and transmembrane proteins that colocalize with the channel core (e.g., Na+ and Ca++ channel β-subunits). For Kv4 channels, evidence of auxiliary subunits initially surfaced when the functional expression and inactivation of Kv4 currents in heterologous cells were modified by some protein factor(s) encoded by low-molecular-weight mRNA transcripts from brain (Chabala et al., 1993, Serodio et al., 1994, Serodio et al., 1996). Therefore, auxiliary proteins that modulate Kv4 channels in heterologous systems may contribute to their function in native tissues.

To discover Kv4 accessory subunits in the brain, the cytoplasmic N-terminal domain (amino acids 1–180) of Kv4.3 was used in yeast two-hybrid (YTH) screens (An et al., 2000). Three related proteins that specifically interact with Kv4 N-terminus were named K+ channel interacting proteins 1–3 (KChIP1, KChIP2, KChIP3) (Table 6). KChIP interaction with Kv4 channels was further substantiated by their coimmunoprecipitation from cotransfected cells and brain tissues as well as their colocalization in vitro and in vivo preparations (An et al., 2000, Holmqvist et al., 2002, Ohya et al., 2001). Northern blot and RT-PCR analysis showed that in rat and mouse, KChIP1 transcripts are predominantly expressed in the brain, KChIP2 transcripts are expressed in the brain and heart, KChIP3 transcripts are mainly expressed in brain with weaker expression in testes, and KChIP4 transcripts are present diffusely in the brain and absent in the heart (An et al., 2000, Holmqvist et al., 2002, Ohya et al., 2001, Spreafico et al., 2001). The distributions of KChIPs have also been investigated by in situ hybridization and immunoflourescence staining, and the results show that KChIP3 mRNA is diffusely present in almost all regions of the brain, with particularly high abundance in hippocampus, olfactory region, cerebral cortex, retina, and cerebellum (Spreafico et al., 2001). At the cellular level, it has been shown that KChIP1 colocalizes with Kv4.3 in hippocampal interneurons and cerebellar granule cells, and KChIP2 colocalizes with Kv4.2 in apical and basal dendrites of hippocampal and neocortical pyrimadal cells and the dendrites of granule cells of dentate gyrus (An et al., 2000). Triple staining of cultured hippocampal interneurons showed that KChIP4, along with KChIP1, showed robust immunofluorescence staining that expectedly colocalized with Kv4.2 in large somatodendritic clusters (Shibata et al., 2003). In short, the combination of coimmunoprecipitation and mRNA/protein colocalization studies demonstrates that KChIPs are constitutive auxiliary subunits of Kv4 channels.

To date, as many as 12 different species of KChIPs have been isolated from the brain and heart through further cDNA cloning, and they are organized based on sequence homology into four main groups (KChIP1-4) with respective splice variants (with alphabetical designations) (Table 6). Surprisingly, KChIP3 is encoded by the same gene locus as two previously characterized proteins, calsenilin and downstream regulatory element antagonist modulator (DREAM) (Spreafico et al., 2001). The KChIP3 sequence is identical to that of calsenilin, a protein that binds presenilin-1 and -2 and regulate proteolytic processing of their C-terminal portion (Buxbaum et al., 1998). Furthermore, KChIP3/calsenilin is also identical to DREAM, except for the absence of 30 residues at the N-terminus (Carrion et al., 1999). Functionally, DREAM has been described as a Ca++-regulated DNA-binding protein that represses transcription of c-fos or prodynorphin genes (Carrion et al., 1999). KChIP4bl, also known as calsenilin-like protein (CALP), is similar to KChIP3 in its ability to bind presenilin-2 (Morohashi et al., 2002).

The known activities of DREAM indicate that KChIPs are Ca++-binding Kv4-interacting proteins. KChIPs share sequence similarities with neuronal calcium sensor (NCS-1) and frequenin and thus are members of the recoverin/NCS subfamily of calcium-binding proteins (Burgoyne and Weiss, 2001). Sequence analysis shows that KChIPs (with the exception of KChIP2d) consist of a variable N-terminus and a highly conserved C-terminal ‘core’ with four EF hand-like motifs (An et al., 2000). The second, third, and fourth EF hands of KChIP strictly conform to the EF-hand consensus, while the first one is less conserved. KChIP1-3 indeed all bind calcium in filter overlay assay and display calcium-dependent mobility shift on SDS-PAGE, confirming that they are calcium-binding proteins (An et al., 2000). Equilibrium binding assay with radiolabeled Ca++ showed that either EF hand 3 or 4 of KChIP3/DREAM strongly bind Ca++ with dissociation constant (Kd) of approximately 0.6 μM (Craig et al., 2002) (Fig. 4), consistent with the fact that Ca++ binding is typically half-maximal below 1 μM for the majority of NCS family members (Burgoyne and Weiss, 2001). Since basal cytosolic Ca++ concentration is around 0.1 μM in mammalian cells, it is possible that the high-affinity Ca++ binding site is partially occupied with Ca++. It is important to note that, although all members of the recoverin/NCS subfamily bind Ca++, only KChIPs possess such robust interactions with Kv4 channels. While frequenin/NCS-1 reportedly interacts with Kv4.2 and Kv4.3 proteins in native tissues and cotransfected cultured cells, their interaction appears to be weaker and/or less effective than the interaction between KChIPs and Kv4 (Nakamura et al., 2001, Guo et al., 2002, Ren et al., 2003). This is consistent with results showing that only a small fraction of frequenin/NCS-1 is associated with Kv4.3 in mouse heart and that frequenin/NCS-1 increases functional expression only slightly (∼2-fold) as compared to KChIPs (∼12-fold) (Nakamura et al., 2001, Guo et al., 2002, An et al., 2000). Other closely related Ca++-binding proteins such as neurocalcin, visinin-like proteins (VILIP)-1, and hippocalcin do not bind Kv4 in vitro (An et al., 2000, Nakamura et al., 2001). These results suggest that interaction with Kv4 channels is due to specific features of KChIPs and is not a general feature of Ca++-binding proteins.

The structural basis of this precise molecular interaction between the cytoplasmic N-terminus of Kv4 channels and KChIPs is beginning to be revealed. Based on the differential ability of KChIP2 and frequenin/NCS-1 to bind Kv4.3 proteins, coimmunoprecipitation analysis using KChIP2-frequenin chimeras shows that two EF-hand linkers (those between first and second EF hands and between third and fourth EF hands), and the C-terminal portion following the fourth EF hand are the determinants of KChIP association with Kv4.3 proteins (Ren et al., 2003). These three structures surround the hydrophobic pocket in the known NCS-1 crystal structure, suggesting that a hydrophobic segment of the Kv4 N-terminus binds with this important hydrophobic pocket. Indeed, the Kv4 cytoplasmic hydrophobic N-terminus is important for the stability of the KChIP structure. Crystals of KChIP1 protein prepared for X-ray crystallography is formed only after the fusion of the first 30 amino acids of the Kv4.2 N-terminus (Kv4.2N) to the critical C-terminal portion of KChIP1 (KChIP1–Kv4.2N) (Zhou et al., 2004). The yielded structure reveals a precise insertion and binding of Kv4.2N with the hydrophobic pocket generated by the first two EF hands and the last α-helix at the C-terminus after the fourth EF hand (Zhou et al., 2004) (Fig. 3, Fig. 4). The Kv4.2N and the last α-helix of KChIP1 allow KChIP dimerization by aligning and forming a dimerization interface in the Ca++-bound structure (Fig. 3, Fig. 4). The dimeric KChIP structure is assembled from calcium-bound KChIP monomers and found in the cytosol, but the precise form of KChIP that associate with Kv4 channels remains unknown. What is known is that, as Kv4 and KChIP associate to form the channel complex, Kv4 subunits appear to assemble with KChIP auxiliary subunits with a 4:4 subunit stoichiometry (Kim et al., 2003). Electron microscopy, in conjunction with single particle reconstruction, of Kv4.2–KChIP2 channels localized the four KChIP moieties laterally juxtaposed to the T1 structures (Kim et al., 2004). Based on these findings, possible scenarios for the Kv4–KChIP complex include a dimer of dimers binding to the tetrameric K+ channel, or four KChIP molecules individually associated with each Kv4 subunit in fourfold symmetric fashion.

While the organization of the functional Kv4–KChIP complex require further clarification, it is known that the binding of Ca++ has important consequences for the structure, oligomerization, and function of KChIP3. In the nucleus, the tetrameric form of KChIP3/DREAM effectively binds to DNA response element (DRE) and represses transcription of the prodynorphin and c-fos genes, and the role of Ca++ is to change KChIP3/DREAM's ability to bind DRE without altering the stability of the DNA-bound tetrameric form (Carrion et al., 1999, Craig et al., 2002, Mellstrom and Naranjo, 2001). By heterologous expression in HEK-293 cells, C-terminal Myc-tagged KChIP3/DREAM is detected in Western blots as monomer, dimer, and tetramer in the nuclear extracts, and as dimer and tetramer in the cytosolic fraction (Carrion et al., 1999). In vitro studies have shown that at low protein concentrations (10–24 μM) the role of Ca++-binding is to dimerize KChIP3/DREAM monomers, as seen in mass spectral, optical spectral, and protein mobility assays (Carrion et al., 1999, Craig et al., 2002, Osawa et al., 2001). However, at higher protein concentrations, tetrameric forms develop at higher protein concentrations irrespective of Ca++-binding (Craig et al., 2002, Osawa et al., 2001).

The molecular basis of how KChIP1, KChIP2, and KChIP3 modify Kv4 expression and gating are beginning to be understood in the context of recent 3D structures. When heterologously expressed alone in COS-1 or CHO cells, Kv4.2 channels are found mostly retained in the perinuclear endoplasmic reticulum, with minimal trafficking to the Golgi compartments and the cell surface (An et al., 2000, Bähring et al., 2001b, Shibata et al., 2003). Deletion of residues 2–40 in the N-terminal domain of Kv4.2 yields increased Kv4.2 currents in HEK-293 cells, and a positive correlation exists between the size of the N-terminal deletion and current enhancement (Bähring et al., 2001b). This suggests that Kv4 channel trafficking is primarily determined by a highly conserved hydrophobic N-terminal sequence. According to the 3D structure, the binding of KChIP masks the ER-retaining N-terminus and allows efficient targeting to the cell surface (Fig. 4). Indeed, KChIPs expressed alone exhibit a diffuse distribution throughout the cytoplasm, where coexpression of KChIP1-3 releases Kv4.2 from its intrinsic ER retention and results in robust trafficking to the Golgi and plasma membrane (An et al., 2000, Bähring et al., 2001b, Shibata et al., 2003).

The physical immobilization of the cytoplasmic N-terminus of Kv4 channels by KChIP directly impacts the fast phase of Kv4 inactivation (Fig. 3). Aside from its novel role in ER retention, the distal hydrophobic N-terminal domain has been established as a participant in open state inactivation of Kv4 channels as described previously. Indeed, the establishment of open-state inactivation and its disengagement through channel reopening and closing are slowed or absent when Kv4 channels are coexpressed with KChIPs (Beck et al., 2002, Boland et al., 2003, Gebauer et al., 2004), and the inactivation time course of Kv4.2 coexpressed with KChIP2.2 resembles that of Kv4.2 N-terminal deletion mutant (Δ2–40) (compare Bähring et al., 2001b with Bähring et al., 2001a). Furthermore, coexpression with KChIP1 remodeled the waveforms of Kv4.1 and Kv4.3 and cause them to become remarkably similar (Beck et al., 2002). Notably, the overall observed macroscopic inactivation of Kv4.1 and Kv4.3 is faster in the presence of KChIP1 (Beck et al., 2002). This indicates that the binding of KChIPs has suppressed the slower phase(s) of Kv4 inactivation, which may result from rearrangements involved in channel closing and closed-state inactivation at an internal site (Jerng et al., 1999). Consistent with this finding, KChIPs accelerated both channel closing and inactivation from the preopen closed states (Beck et al., 2002, Boland et al., 2003). Furthermore, mutations in the inner vestibule of Kv4.3 (V(399,401)I) significantly increased the sustained, noninactivating component and eliminated the ability of KChIP2b to accelerate slow phase of inactivation (Wang et al., 2002).

While the conserved C-terminal “core” of KChIPs binds calcium and contains the structures necessary for modulation of Kv4 expression and kinetics, precise functions attributed to the variable N-terminal domain is only beginning to be elucidated. In one example, a unique splice variant of the KChIP4 gene (KChIP4a) prominently fails to increase surface expression but dramatically slows inactivation of Kv4.1 and Kv4.3 channels in Xenopus oocytes and CHO cells with modest changes in other channel properties (Holmqvist et al., 2002). The N-terminal domain of KChIP4a, with its long stretch of hydrophobic residues, underlies these exceptional effects and was thus appropriately named the K+-channel inactivation suppressor (KIS). In another example, KChIP1b possesses the unusual property of slowing the kinetics of recovery from inactivation in Kv4 channels (Van Hoorick et al., 2003). KChIP1b is a longer splice variant of KChIP1a, with an additional exon that introduces a novel 11-residue-long sequence (DIAWWYYQYQR) into KChIP1a's N-terminal domain (Boland et al., 2003, Van Hoorick et al., 2003). Similar to the KIS domain of KChIP4a, the extra exon is rich in amino acids with aromatic side chains that are predicted to arrange in an α-helical conformation (Van Hoorick et al., 2003).

How does the KIS domain prevent the expression augmenting effects of KChIP “core”? Unlike other KChIPs, coexpression of KChIP4a with Kv4.2 does not result in increased trafficking of the channel to the surface, consistent with a lack of increased current expression (Shibata et al., 2003). Where KChIP1-3 shows diffuse cytoplasmic distribution, KChIP4a immunostaining in COS-1 is perinuclear, suggesting an association with the ER (Shibata et al., 2003). It appears that, much like the Kv4 N-terminus, the KIS domain may also act to retain the protein complex in the ER and interfere with surface trafficking. How does the KIS domain abolish fast inactivation of Kv4 channels? Although it has not been directly demonstrated, KChIP4a possesses the conserved C-terminal domain involved in Kv4 association and thus may also remove open state inactivation by binding to the Kv4 N-terminus (Fig. 3). In addition to delaying activation, KChIP4a slows the rate of deactivation and therefore may indirectly delay preopen closed-state inactivation by mechanisms reviewed before (Holmqvist et al., 2002). Overall, the KIS domain may keep the channel open by interfering with fast and slow processes that control inactivation at internal sites. Meanwhile, the presence of the KIS domain prevents the KChIP C-terminal domain's ability to cause shifts in the voltage dependence of steady-state inactivation and acceleration of recovery from inactivation (Holmqvist et al., 2002). In summary, when KChIP4a is bound to the Kv4 N-terminus, the hydrophobic KIS domain acts to retain the channel in the ER, interfere with inactivation, and prevent acceleration of recovery from inactivation.

KChIPs are integral modulatory components of Kv4 channels that reconstitute some of the properties of ISA from mammalian central neurons. Although some KChIPs induce an overall faster development of inactivation with a slower initial decay (Beck et al., 2002), other biophysical properties were not remodeled to match the functional properties of ISA. For instance, KChIPs generally shift the voltage dependence of steady-state prepulse inactivation toward more positive voltages as the result of a faster recovery from inactivation that dominates the overall equilibrium of closed-state inactivation (Beck et al., 2002). In contrast, it is common to observe ISA with an overall fast development of inactivation and a hyperpolarized voltage dependence of prepulse inactivation (Table 2), which suggested the existence of additional remodeling or modulatory components. High-molecular-weight mRNAs (4–7 kb) from rat cerebellum encodes a molecular factor that accelerates the kinetics of Kv4 channels expressed in Xenopus oocytes, and the activity of this factor cannot be eliminated by antisense hybrid suppression of KChIP expression (Nadal et al., 2001). The identity of this Kv4 modulator was revealed when immunoprecipitation of Kv4 channels from rat cerebellar membranes yielded a 115-kDa integral membrane protein named dipeptidyl aminopeptidase-like protein (DPPX, also known as DPP6 or BSPL) (Nadal et al., 2003). Sequence analysis shows that DPPX is a member of the dipeptidyl aminopeptidase family, with approximately 30% identity and 50% similarity with reported yeast and rat dipeptidyl aminopeptidases, DPP4 or CD26 (Wada et al., 1992). DPPX protein has a short N-terminal cytoplasmic domain, a single transmembrane segment, and a large C-terminal extracellular domain. Two adult forms, DPPX-L (long) and DPPX-S (short), and one embryonic form of DPPX are known and differ mainly in the length and composition of their N-terminal cytoplasmic domains (Hough et al., 1998, Wada et al., 1992) (Table 7). By analogy to DPP4/CD26, the C-terminal extracellular domain can be tentatively divided into an N-glycosylation-rich region, an extracellular matrix-binding domain (cys-rich), and an aminopeptidase domain. While the three-dimensional structure of DPPX is yet unsolved, X-ray crystallography, cryo-TEM, and single particle analysis determine that the extracellular region of the related DPP4/CD26 consists of a unique eight-bladed β-propeller domain and a serine protease domain (Hiramatsu et al., 2003, Ludwig et al., 2003, Oefner et al., 2003, Rasmussen et al., 2003). As part of the complete DPP4 protein, the serine protease domain would be externally juxaposed to the cell membrane, with the β-propeller domain farther off into the milieu.

DPPX-S and DPPX-L polypeptides are integral membrane glycoproteins initially synthesized with high mannose sugars but later linked with complex sugars at their seven N-glycosylation sites distributed throughout the three extracellular domains (Kin et al., 2001). Although in DPP4/CD26 the catalytic domain contains serine protease activity that cleaves N-terminal dipeptides when the second residue is proline (Axx-Pro), the catalytic peptidase domain in DPPX is inactive: the catalytic serine at the active site (Gly–Trp–Ser–Tyr–Gly) is substitued by aspartate, but a reversion of this residue fails to restore activity (Kin et al., 2001). DPPX is widely expressed in the brain, with some minor expressions in kidneys, ovaries, and testis. In situ hybridization for DPPX-L mRNA signals the presence of its message in the dendritic processes of hippocampal pyramidal neurons, and the steady-state level of the message increases following tetanic stimultation to produce LTP (de Lecea et al., 1994). DPPX's dendritic distribution is corroborated by its coimmunoprecipitation with Kv4.2 proteins in complex with KChIP in rat cerebellar membranes (Nadal et al., 2003).

Coexpression studies in Xenopus oocytes and CHO cells show that DPPX-S dramatically facilitated channel surface trafficking and conferred many features of the native ISA channels by altering the functional properties of Kv4 channels (Nadal et al., 2003). The association of DPPX with Kv4 channels increases current expression between 3 and 25 times due to increased trafficking, as determined by current recordings and immunohistochemistry. For Kv4.2 and Kv4.3, DPPX-S shifts the voltage dependence of activation and inactivation to more hyperpolarized potentials and significantly accelerates the time courses of activation, inactivation, and recovery from inactivation (Nadal et al., 2003). The rate of closed-state inactivation at the V1/2 membrane potential is also increased dramatically by as much as 10- to 15-fold (Jerng et al., 2004, Rocha et al., 2004). The leftward shift in steady-state inactivation of Kv4.1 and Kv4.2 channels may be due to the dominance of entry into closed-state inactivation over recovery from inactivation. Thus, the overall effect of DPPX is to enhance preferential closed-state inactivation.

The molecular composition of the Kv4-associated complex underlying neuronal ISA is further elaborated by suggestions of additional protein factors that primarily act as chaperones, assisting Kv4 channels in their trafficking to the plasma membrane or their turnover rate once there. In general, these accessory factors do not exhibit the ability to alter channel properties, and they have not been as well studied as the modulatory auxiliary subunits. In the brain, the reported Kv4 accessory factors include Kvβ, postsynaptic density 95 (PSD-95), and filamin. Kv1-type subunits (Kvβ) are critical auxiliary subunits that bind to the hydrophilic N-terminus of Kv1 channels, act as molecular chaperones, and may confer fast N-type inactivation to noninactivating channels (Rettig et al., 1994). Confounding reports have suggested that, while they have no effects on Kv4.3 channel gating, Kv1 accessory subunits (Kvβ1 and Kvβ2) interact with Kv4.3 channels in rat brain and mildly enhances their surface expression by less than twofold in heterologous systems by interacting with the cytoplasmic C-terminus, which was later changed to be the N-terminus (Wang et al., 2003, Yang et al., 2001). It has also been proposed that Kvβ subunits may be physiologically important since Kvβ1.2 confers O2 sensitivity to Kv4.2 but not to Shaker channels when coexpressed in HEK-293 cells (Perez-Garcia et al., 1999).

Ion channels and receptors are membrane-bound proteins directed to the surface and subsequently supported by cytoskeletal proteins; therefore, their association with cytoskeletal proteins may increase the residency time in the membrane. PSD-95 proteins contain PDZ domains that bind to the C-termini of Kv1.1, Kv1.2, and Kv1.4 channels (Kim et al., 1995), and coexpression of these Kv1 channels with PSD-95 in heterologous cells results in high density plasma membrane clusters (Kim and Sheng, 1996, Tiffany et al., 2000). Coimmunoprecipitation of Kv4.2 and PSD-95 coexpressed in mammalian cell line suggests that they too associate, and this association occurs through a novel C-terminal valine–serine–alanine–leucine (VSAL) motif (Wong et al., 2002). As with Kv1 channels, PSD-95 increased the surface expression of Kv4.2 channels by approximately twofold and induced channel clustering. However, the role of PSD-95 in K channel clustering is called into question since Kv1 and the neurexin Caspr2, both containing the C-terminal PDZ-binding motifs, still clusters at juxtaparanodes in a mutant mouse lacking juxtaparanodal PSD-95 (Rasband et al., 2002). Filamin, another cytoskeletal structural protein, directly binds to actin and Kv4.2 channels at a proline-rich region (PTPP) in the cytoplasmic C-terminal region and increases current density approximately 2.7-fold (Petrecca et al., 2000). Therefore, filamin may act as an intermediary between Kv4.2 and the cytoskeletal actin in the postsynaptic density.

Section snippets

Modulation of ISA by phosphorylation: Ca++/calmodulin-dependent protein kinase II (Ca++MKII), cAMP-dependent protein kinase (PKA), protein kinase C (PKC), and mitogen-actived protein kinase (MAPK)

The neuronal ISA supermolecular complexes formed from Kv4 α-subunits, KChIP/DPPX β-subunits, and accessory proteins are regulated by the activity of protein kinases. The various protein kinase pathways that have been shown to produce phosphorylation of the ISA/Kv4 complex in hippocampal CA1 neurons are illustrated in Fig. 5A. In this section, we will examine the modulation of expression and function that are produced by posttranslational modification of ISA/Kv4 channels and auxiliary subunit.

Acknowledgments

We thank Drs. Aguan Wei (Washington University, St. Louis, MO) and Daniel Johnston (Baylor College of Medicine, Houston, TX) for critically reading this manuscript. In addition, we thank Debra S. Karhson for her valuble assistance in Fig. 2 and former and current members of the Covarrubias' lab for their contributions and feedback. The preparation of this manuscript was supported by a research grant from the National Institute of Neurological Disorders and Stroke, NIH (R01 NS32337 to M.C.); the

References (246)

  • M. Covarrubias et al.

    Elimination of rapid potassium channel inactivation by phosphorylation of the inactivation gate

    Neuron

    (1994)
  • T.A. Craig et al.

    The metal-binding properties of DREAM: evidence for calcium-mediated changes in DREAM structure

    J. Biol. Chem.

    (2002)
  • R.J. Davis

    The mitogen-activated protein kinase signal transduction pathway

    J. Biol. Chem.

    (1993)
  • L. de Lecea et al.

    Transcripts encoding a neural membrane CD26 peptidase-like protein are stimulated by synaptic activity

    Brain Res. Mol. Brain Res.

    (1994)
  • D. del Camino et al.

    Tight steric closure at the intracellular activation gate of a voltage-gated K+ channel

    Neuron

    (2001)
  • S.D. Demo et al.

    The inactivation gate of the Shaker K+ channel behaves like an open-channel blocker

    Neuron

    (1991)
  • L. Dryer et al.

    Arachidonic acid-sensitive A-currents and multiple Kv4 transcripts are expressed in chick ciliary ganglion neurons

    Brain Res.

    (1998)
  • F.R. Fernandez et al.

    Inactivation of Kv3.3 potassium channels in heterologous expression systems

    J. Biol. Chem.

    (2003)
  • M. Gebauer et al.

    N-type inactivation features of Kv4.2 channel gating

    Biophys. J.

    (2004)
  • F. Gomez-Lagunas et al.

    The relation between ion permeation and recovery from inactivation of ShakerB K+ channels

    Biophys. J.

    (1994)
  • W. Guo et al.

    Modulation of Kv4-encoded K+ currents in the mammalian myocardium by neuronal calcium sensor-1

    J. Biol. Chem.

    (2002)
  • K. Hagiwara et al.

    Differential inhibition of transient outward currents of Kv1.4 and Kv4.3 by endothelin

    Biochem. Biophys. Res. Commun.

    (2003)
  • S. Heinemann et al.

    The inactivation behaviour of voltage-gated K-channels may be determined by association of α- and β-subunits

    J. Physiol. (Paris)

    (1994)
  • S.H. Heinemann et al.

    Molecular and functional characterization of a rat brain Kvβ3 potassium channel subunit

    FEBS Lett.

    (1995)
  • H. Hiramatsu et al.

    The structure and function of human dipeptidyl peptidase IV, possessing a unique eight-bladed β-propeller fold

    Biochem. Biophys. Res. Commun.

    (2003)
  • R. Horn

    A new twist in the saga of charge movement in voltage-dependent ion channels

    Neuron

    (2000)
  • T. Hoshi et al.

    Two types of inactivation in Shaker K+ channels: effects of alterations in the carboxy-terminal region

    Neuron

    (1991)
  • C.F. Hsiao et al.

    Characteristics of a fast transient outward current in guinea pig trigeminal motoneurons

    Brain Res.

    (1995)
  • D. Isbrandt et al.

    Gene structures and expression profiles of three human KCND (Kv4) potassium channels mediating A-type currents ITO and ISA

    Genomics

    (2000)
  • L.L. Isom et al.

    Auxiliary subunits of voltage-gated ion channels

    Neuron

    (1994)
  • A.W. Jahng et al.

    Zinc mediates assembly of the T1 domain of the voltage-gated K channel 4.2

    J. Biol. Chem.

    (2002)
  • H.H. Jerng et al.

    K+ channel inactivation mediated by the concerted action of the cytoplasmic N- and C-terminal domains

    Biophys. J.

    (1997)
  • D.C. Johns et al.

    Suppression of neuronal and cardiac transient outward currents by viral gene transfer of dominant-negative Kv4.2 constructs

    J. Biol. Chem.

    (1997)
  • J.P. Adams et al.

    The A-type potassium channel Kv4.2 is a substrate for the mitogen-activated protein kinase ERK

    J. Neurochem.

    (2000)
  • R.W. Aldrich

    Fifty years of inactivation

    Nature

    (2001)
  • R.W. Aldrich et al.

    Inactivation of delayed outward current in molluscan neurone somata

    J. Physiol.

    (1979)
  • R.W. Aldrich et al.

    A reinterpretation of mammalian sodium channel gating based on single channel recording

    Nature

    (1983)
  • R.W. Aldrich et al.

    Differences in gating among amino-terminal variants of Shaker potassium channels

    Cold Spring Harbor Symp. Quant. Biol.

    (1990)
  • G. Alonso et al.

    Clustering of Kv4.2 potassium channels in postsynaptic membrane of rat supraoptic neurons: an ultrastructural study

    Neuroscience

    (1997)
  • W.F. An et al.

    Modulation of A-type potassium channels by a family of calcium sensors

    Nature

    (2000)
  • C. Antz et al.

    NMR structure of inactivation gates from mammalian voltage-dependent potassium channels

    Nature

    (1997)
  • R.K. Ayer et al.

    Enhanced closed-state inactivation in a mutant Shaker K+ channel

    J. Membr. Biol.

    (1997)
  • R. Bähring et al.

    Kinetic analysis of open- and closed-state inactivation transitions in human Kv4.2 A-type potassium channels

    J. Physiol.

    (2001)
  • M.I. Banks et al.

    Layer-specific properties of the transient K current (IA) in piriform cortex

    J. Neurosci.

    (1996)
  • R. Bardoni et al.

    Kinetic study and numerical reconstruction of A-type current in granule cells of rat cerebellar slices

    J. Neurophysiol.

    (1993)
  • D.J. Baro et al.

    Lobster Shal: comparison with Drosophila Shal and native potassium currents in identified neurons

    J. Neurosci.

    (1996)
  • T. Baukrowitz et al.

    Use-dependent blockers and exit rate of the last ion from the multi-ion pore of a K+ channel

    Science

    (1996)
  • E.J. Beck et al.

    Interactions between multiple phosphorylation sites in the inactivation particle of a K+ channel. Insights into the molecular mechanism of protein kinase C action

    J. Gen. Physiol.

    (1998)
  • E.J. Beck et al.

    Remodelling inactivation gating of Kv4 channels by KChIP-1, a small-molecular-weight calcium binding protein

    J. Physiol.

    (2002)
  • J.M. Bekkers

    Distribution and activation of voltage-gated potassium channels in cell-attached and outside-out patches from large layer 5 cortical pyramidal neurons of the rat

    J. Physiol.

    (2000)
  • Cited by (0)

    View full text