Structure and expression of principal and auxiliary subunits of native Kv channels in mammalian brain
Kv channels are the most diverse subfamily of voltage-gated ion channel α subunits, with >40 human genes in 12 subfamilies, termed Kv1–Kv12 (Yu & Catterall, 2004). Kv channel α subunits have six transmembrane segments, the first four (S1–S4) form the voltage-sensing module, and the region between transmembrane segments S5–S6 forms the pore module (Fig. 1). Four independent α subunits from the same subfamily assemble to form functional homomeric or heteromeric Kv channels, and coassembly with auxiliary subunits further enhances Kv channel diversity. Subunit composition determines the trafficking, localization, biophysical, pharmacologic, and modulation characteristics of a given Kv channel.
Figure 1. Cartoon showing transmembrane topology of a prototypical Kv1 (left), Kv2 (middle), and Kv7 (right) channel α subunit, with the voltage-sensing module comprising transmembrane segments S1–S4 (positive charges shown in S4) to the left, and the pore module comprising S5–S6 to the right. Colored rectangles show approximate locations of identified trafficking determinants or TDs. Colored ovals show approximate binding sites of interacting proteins implicated in trafficking to and/or scaffolding at the AIS. Yellow circles denote specific sites of epilepsy-associated mutations known to impact trafficking.
Download figure to PowerPoint
In general, Kv channels exhibit subfamily-specific patterns of subcellular localization (Vacher et al., 2008), with Kv1, Kv2, and Kv7 family members unique in that they are found at the AIS (Fig. 2) (Vacher et al., 2008; Clark et al., 2009). Kv1 channels at the AIS play crucial roles in determining action potential initiation and propagation (Kole & Stuart, 2012). These channels generally contain Kv1.1 and Kv1.2 α subunits and auxiliary Kvβ2 subunits (Van Wart et al., 2007; Lorincz & Nusser, 2008; Ogawa et al., 2010; Duflocq et al., 2011; Vacher et al., 2011), although Kv1.4 is found at the AIS in younger animals (Ogawa et al., 2008). Delayed rectifier-type Kv2.1 and Kv2.2 channels (Lim et al., 2000; Johnston et al., 2008; Sarmiere et al., 2008) and M-type Kv7.2 and Kv7.3 channels (Cooper et al., 2001; Cooper, 2011; Klinger et al., 2011) are present at high density at the AIS. Other Kv channel subunits have not been described at the AIS (Vacher et al., 2008).
Figure 2. Localization of Kv1 α subunits at the AIS in the mammalian nervous system. Top left panels: Nav1.6 (green) and Kv1.2 (red) α subunits in a layer 5 pyramidal neuron in rat neocortex (Lorincz & Nusser, 2008). Bottom left panel: Kv2.1 (green) α subunits and Ank-G (red) in neuron within the CA1 region of rat hippocampus (Sarmiere et al., 2008). Right panels. KCNQ2 (red) and KCNQ3 (green) α subunits in the axon initial segment of a mouse spinal cord ventral horn motoneuron; blue is the nucleic acid stain 4′, 6-diamidino-2-phenylindole (Pan et al., 2006).
Download figure to PowerPoint
Kv channel trafficking and clustering determinants and mechanisms
Specific amino acid sequence motifs acting as trafficking determinants (TDs) have been identified within Kv1, Kv2, Kv7α, and Kvβ auxiliary subunits, and direct the establishment and maintenance of expression and localization of these channels at the AIS. Many of these TDs are located within the intrinsically disordered C-terminal domains, which are thought to act as “intermolecular fishing rods” for interacting proteins important in Kv-channel trafficking (Magidovich et al., 2006). These C-terminal domains are highly conserved in mammalian channel orthologs (e.g., human and rat Kv1.2 are 100% identical in the cytoplasmic C-terminus and 99% identical overall), yet are the most divergent regions within the α subunits in mammalian paralogs (e.g., human Kv2.1 and Kv2.2 are only 39% identical in the C-terminus, yet are otherwise 92% identical, and 62% identical overall).
A number of TDs regulate biogenic trafficking of Kv channels through the endomembrane system, from their sites of translation in the endoplasmic reticulum (ER), to their sites of action in the plasma membrane (Fig. 1). These include an anterograde TD sequence VXXSL present within the Kv1.4 cytoplasmic C-terminus, but not in other Kv1 family members (Li et al., 2000), and a potent ER retention TD in the turret domain external to the Kv1.1 channel pore that is dominant to the VXXSL TD (Manganas et al., 2001b), and which acts to dictate the subunit composition-dependent trafficking of heteromeric Kv1 channels (Manganas & Trimmer, 2000). These TDs can act interdependently to regulate trafficking and plasma membrane expression of Kv1 channels (Zhu et al., 2003).
The same amino acids that form the potent turret TD confer sensitivity to dendrotoxin, or DTX, which suggests that a DTX-like molecule within the ER could mediate ER retention by binding to this TD (Manganas et al., 2001b). Expression of soluble DTX within the ER lumen enhances cell surface expression of homomeric and heteromeric channels containing wild-type DTX-sensitive Kv1.1, but not those containing Kv1.1 mutants with reduced DTX binding (Vacher et al., 2007). The matrix metalloprotease MMP23 has a domain similar to a toxin with a Kv1 channel binding selectivity distinct from DTX. Overexpression of MMP23 leads to intracellular retention of Kv1 channels in order of the strength of their MMP23 binding (Rangaraju et al., 2010), suggesting that other MMP23-like mammalian proteins that contain a DTX-like domain could mediate intracellular trafficking by binding to the turret TD. Other Kv1 channel TDs (Fig. 1) include an acidic motif in a membrane-proximal region of the cytoplasmic C-terminus of Kv1 α subunits (Manganas et al., 2001a), and a specific motif within the extracellular loop between transmembrane segments S1 and S2 (McKeown et al., 2008), although the impact of altering these TDs is likely due to folding defects and ER quality control (Braakman & Bulleid, 2011).
The Kv1.2 α subunit contains weak versions of these TDs and, as such, has trafficking characteristics sensitive to modulation by coassembly with Kv1 α subunits containing stronger TDs (Manganas & Trimmer, 2000) and with auxiliary Kvβ subunits (Shi et al., 1996; Campomanes et al., 2002; Gu et al., 2003). Phosphorylation of Kv1.2 also regulates its trafficking, for example upon acute suppression of Kv1.2 current upon muscarinic stimulation (Huang et al., 1993), mediated by endocytosis triggered by phosphorylation of specific C-terminal tyrosine residues (Nesti et al., 2004), and leading to disruption of Kv1.2 interaction with the cytoskeletal protein cortactin (Williams et al., 2007). Note that phosphorylation at a different C-terminal tyrosine residue regulates Kv1.2 clustering (Gu & Gu, 2011; Smith et al., 2012). C-terminal serine phosphorylation sites regulate biogenic trafficking of Kv1.2-containing channels (Yang et al., 2007), and protein kinase A-mediated enhancement of Kv1.2 currents (Johnson et al., 2009).
Polarized trafficking of Kv1 channels to axons is dependent upon the T1 domain within the cytoplasmic N-terminus (Rivera et al., 2005) that serves as a binding site for Kif5b subunits of the microtubule motor kinesin 1 (Rivera et al., 2007). Another kinesin subunit, KIF3A, a component of kinesin 2, is involved in polarized trafficking of Kv1.2 channels to axons (Gu et al., 2006; Gu & Gu, 2010). Kinesin-based trafficking of Kv1 channels is strongly influenced by cytoplasmic Kvβ auxiliary subunits, which also bind to the T1 domain. All three Kvβ subunit isoforms (Kvβ1–3) can promote cell surface expression of associated Kv1 channel complexes via effects on biogenic ER export (Shi et al., 1996; Campomanes et al., 2002), whereas coexpression of Kvβ2 with intact Kv1.2 (Campomanes et al., 2002) or reporter constructs containing the Kv1.2 T1 domain (Gu et al., 2003) increases their polarized expression in axons versus dendrites. Kvβ2 interacts directly with the microtubule-binding protein EB1 (Gu et al., 2006; Vacher et al., 2011), and knockdown of EB1 suppresses Kv1.2 trafficking to axons (Gu & Gu, 2010). Phosphorylation of Kvβ2 regulates its interaction with EB1, and alters trafficking of Kv1 channels to axons and specifically to the AIS (Vacher et al., 2011), suggesting a mechanism for the dynamic regulation of ion channel localization at the AIS (Kuba et al., 2006, 2010; Grubb & Burrone, 2010a; Grubb et al., 2011). Kv1 α subunits contain a PDZ-binding motif (Fig. 1) on their C-termini (Kim et al., 1995), and specific PDZ domain-containing proteins of the membrane-associated guanylate kinase family, specifically PSD-93, are found associated and colocalized with Kv1 channels at the AIS (Ogawa et al., 2008). The precise targeting mechanisms that generate the observed colocalization of these proteins to the AIS are not known.
Unlike Kv1 channels, biogenic intracellular trafficking of Kv2.1 channels to the plasma membrane is not strongly influenced by TDs (Shi et al., 1994; Lim et al., 2000). However, increased insertion of Kv2.1 into the plasma membrane occurs acutely in response to stimuli that induce neuronal apoptosis (Yu et al., 1999; Pal et al., 2003, 2006; Redman et al., 2007) via increased Kv2.1 phosphorylation at specific C-terminal sites (Redman et al., 2007), and in response to monocular deprivation, leading to changes in neuronal excitability and intrinsic plasticity in visual cortical neurons (Nataraj et al., 2010). Changes in Kv2.1 phosphorylation at other C-terminal sites regulate whether Kv2.1 is clustered or is dispersed across the somatodendritic membrane (Misonou et al., 2004, 2005, 2006), although Kv2.1 channels at the AIS of cultured neurons are somewhat refractory to dephosphorylation-dependent dispersal (Misonou et al., 2004). The clustering of Kv2.1 in neurons is mediated by a specific C-terminal domain (the proximal restriction and clustering domain, Fig. 1) that also restricts the localization of clustered Kv2.1 to proximal dendrites, and to the AIS (Lim et al., 2000). Phosphorylation of Kv2.1 by cyclin-dependent kinase 5 is required for maintenance of AIS clustering (Cerda & Trimmer, 2011). Recent studies suggest that Kv2.2 exists in somatodendritic clusters in certain brain neurons (Hermanstyne et al., 2010), in some cases associated with Kv2.1 (Kihira et al., 2010). Kv2.2 has been proposed to act at the AIS of auditory neurons (Johnston et al., 2008). The precise mechanism of Kv2 channel clustering at the AIS has not been determined, although compelling data have been provided supporting a role for a perimeter fence–based mechanism underlying the somatodendritic clustering of Kv2.1 (O’Connell et al., 2006; Tamkun et al., 2007).
Neuronal M-type Kv7 channels at the AIS exist as heteromers of Kv7.2 and Kv7.3 α subunits (Fig. 2) (Cooper, 2011). Coassembly of Kv7.2 and Kv7.3 is crucial to their efficient trafficking, as the respective homomeric channels exhibit poor cell surface expression due to ER retention (Schwake et al., 2003; Rasmussen et al., 2007). Heteromeric Kv7.2/Kv7.3 assembly is also required for efficient localization at the AIS (Rasmussen et al., 2007). Trafficking of homomeric Kv7.3 is especially deficient, a characteristic that has been related to a Kv7.3-specific Ala at position 315 (as opposed to a Thr at this position in all other Kv channels) that acts as a potent pore TD, such that its mutation to Thr allows for expression of homomeric Kv7.3 channels (Gomez-Posada et al., 2010) (Fig. 1). The location of this residue deep within the channel pore makes it unlikely that retention is mediated by interaction with a luminal ER protein, and more likely due to misfolding of homomeric Kv7.3 channels and retention via ER quality control (Braakman & Bulleid, 2011).
Numerous protein–protein interactions are mediated by the C-terminus of Kv7 α subunits (Haitin & Attali, 2008) (Fig. 1). This includes interaction of Kv7.2 and Kv7.3 with ankyrin-G (Ank-G), via a C-terminal binding motif, which localizes Kv7 channels to the AIS (Cooper, 2011), and of Kv7.2 with calmodulin (CaM), which plays a particularly critical role in trafficking (Etxeberria et al., 2008; Alaimo et al., 2009). The C-terminus of Kv7 α subunits is also extensively modified by phosphorylation, with proteomic analyses identifying at least 14 in vivo sites in Kv7.2, and four sites in Kv7.3 (Cerda et al., 2011), providing a potential mechanism for dynamic regulation of Kv7 channel trafficking. Trafficking of Kv7 channels is dynamically regulated in epithelial cells, with intracellular retention in unpolarized cells, and plasma membrane expression upon establishment of epithelial polarity (Andersen et al., 2011).
Defects in Kv trafficking in epilepsy
Among the Kv1 and Kv2 α and Kvβ subunits known to be associated with the AIS, only the KCNA1 gene (http://omim.org/entry/176260) encoding the Kv1.1 α subunit has mutant alleles associated with neurologic disorders, specifically episodic ataxia type 1 or EA-1, often associated with seizures, which arises due to loss of function of Kv1.1-containing channels. Numerous EA-1 missense mutations have been identified that reduce or eliminate trafficking of Kv1.1-containing channels, as well as those that affect channel gating (Kullmann, 2010). A truncated mutant generated from a specific EA-1 nonsense mutation “RX” (Rea et al., 2002) lacks the membrane proximal acidic TD in the Kv1.1 C-terminus (Fig. 1), causing misfolding and intracellular retention of the mutated Kv1.1 α subunits (Manganas et al., 2001a). It is notable that co-assembly of these mutated RX subunits with otherwise normal Kv1 α subunits acts as a dominant negative for their folding and trafficking (Manganas et al., 2001a; Rea et al., 2002), and expression of RX in neurons leads to reduced Kv current levels, and increased glutamate release, consistent with loss of expression of RX-containing channels, including those at the AIS (Heeroma et al., 2009), and providing a possible basis for the severe EA-1 symptoms associated with this mutation (Eunson et al., 2000).
The KCNA2 gene encoding Kv1.2 is not known to harbor mutations resulting in neurologic or other human diseases (http://omim.org/entry/176262); however, Kv1.2 knockout (KO) mice have enhanced susceptibility to flurothyl-induced seizures at postnatal day 14 (P14), at P15 begin to have spontaneous running-bouncing seizures with tonic extension, and by P19 are dead from seizures (Brew et al., 2007). A mouse ENU-induced mutant that exhibits chronic motor uncoordination, named Pingu, carries a KCNA2 missense mutation yielding an I402T mutation in the S6 transmembrane segment, resulting in decreased Kv1.1 and Kv1.2 expression in cerebellum, and heterologous cells expressing this mutant have decreased expression, suggesting a trafficking defect (Xie et al., 2010). The KCNAB2 gene encoding Kvβ2 does not harbor disease-causing mutations (http://omim.org/entry/601142), but certain patients with 1p36 deletion syndrome lack the chromosomal region containing the Kvβ2 gene (Shapira et al., 1997), and there exists a strong association between seizure severity and deletion of KCNAB2 (Heilstedt et al., 2001; Kurosawa et al., 2005). Disease-associated mutations in Kv2.1 and Kv2.2 (http://omim.org/entry 600397 and 607738, respectively) have not been described.
Eleven distinct allelic variants of Kv7.2 are associated with benign familial neonatal epilepsy (BFNE) (http://omim.org/entry/602235), as are mutations in Kv7.3 (http://omim.org/entry/602232). Kv7 channels are also important as targets for retigabine, a novel antiepileptic drug that acts as a positive allosteric modulator in promoting the enhanced opening of neuronal Kv7 channels (Gunthorpe et al., 2012). One set of BFNE-associated mutations clusters in the Kv7.3 primary sequence just upstream (D305G, W309R, G310V) and downstream (R330C) of the critical pore Ala-315 TD (Fig. 1). It is not known whether these mutants exhibit reduced current due to reduced single-channel conductance or reduced expression (Maljevic et al., 2010), although the G310V mutant has reduced levels of axonal surface expression in cultured neurons (Chung et al., 2006). A mutation in the pore region of Kv7.2 results in misfolding and reduced trafficking; however the channels that appear at the cell surface have normal single-channel conductance (Maljevic et al., 2011). It is intriguing that a large number of BFNE mutations cluster within the Kv7.2 C-terminus, including missense, nonsense, and frame-shift mutations. Not surprisingly, many of these yield reduced expression via effects on trafficking of homomeric and heteromeric Kv7.2-containing channels (Maljevic et al., 2010). Some of these (Fig. 1) appear to act through decreasing binding of CaM (Richards et al., 2004) such as Kv7.2 mutations R353G (Etxeberria et al., 2008) and L339R (Alaimo et al., 2009), which impact the exit of Kv7.2-containing channels from the ER, as well as L619R (Richards et al., 2004), the trafficking properties of which have not been determined. Although the role of CaM in Kv7-channel function is complex (Haitin & Attali, 2008), and includes effects on gating and binding to other interacting proteins such as A-kinase anchor proteins (Bal et al., 2010), CaM binding, and its disruption by mutations found in epilepsy, impact trafficking. BFNE mutations have also been found to alter the polarized localization and plasma membrane expression of Kv7.2/Kv7.3 channels in axons (Chung et al., 2006).