Subunit Composition of Kv1 Channels in Human CNS

Authors

  • Sarah K. Coleman,

  • Jia Newcombe,

  • Jonathan Pryke,

  • J. Oliver Dolly


  • Abbreviations used : PMSF, phenylmethylsulphonyl fluoride ; SDS-PAGE, sodium dodecyl sulfate-polyacrylamide gel electrophoresis ; TES, N-tris(hydroxymethyl)methyl-2-aminoethanesulphonic acid.

Address correspondence and reprint requests to Dr. J. O. Dolly at Department of Biochemistry, Imperial College of Science, Technology and Medicine, London, SW7 2AZ, U.K.

Abstract

Abstract : The α subunits of Shaker-related K+ channels (Kv1.X) show characteristic distributions in mammalian brain and restricted coassembly. Despite the functional importance of these voltage-sensitive K+ channels and involvement in a number of diseases, little progress has been achieved in deciphering the subunit composition of the (α)4(β)4 oligomers occurring in human CNS. Thus, the association of α and β subunits was investigated in cerebral grey and white matter and spinal cord from autopsy samples. Immunoblotting established the presence of Kv1.1, 1.2, and 1.4 in all the tissues, with varying abundance. Sequential immunoprecipitations identified the subunits coassembled. A putative tetramer of Kv1.3/1.4/1.1/1.2 was found in grey matter. Both cerebral white matter and spinal cord contained the heterooligomers Kv1.1/1.4 and Kv1.1/1.2, similar to grey matter, but both lacked Kv1.3 and the Kv1.4/1.2 combination. An apparent Kv1.4 homooligomer was detected in all the samples, whereas only the brain tissue possessed a putative Kv1.2 homomer. In grey matter, Kvβ2.1 was coassociated with the Kv1.1/1.2 combination and Kv1.2 homooligomer. In white matter, Kvβ2.1 was associated with Kv1.2 only, whereas Kvβ1.1 coprecipitated with all the α subunits present. This represents the first description of Kv1 subunit complexes in the human CNS and demonstrates regional variations, indicative of functional specialisation.

Voltage-gated Kv channels are a large and very diverse group, consisting of multiple subfamilies (Kv1-4). They are extremely important for normal functioning of the nervous system, contributing to the control of spike frequency, modulation of neurotransmitter release, and setting of resting potential threshold. Kv1 Shaker-related channels from mammalian brain have been purified by the use of selective inhibitors, such as dendrotoxins from the venom of Dendroaspis snakes (Dolly et al., 1984 ; Rehm and Lazdunski, 1988 ; Parcej and Dolly, 1989). The resultant preparations were shown to be octomeric sialoglycoproteins consisting of four α subunits, which are transmembrane, pore-forming glycoproteins of Mr = 58,000-95,000, and four β subunits, which are cytoplasmic auxiliary proteins of Mr ~ 39,000 (Scott et al., 1990 ; Parcej et al., 1992). Seven separate genes encoding α-subunit variants, Kv1.1-1.7 (Gutman and Chandy, 1993), have been studied extensively, and two additional members were reported recently. Since the initial discovery of the Kvβ1 and β2 genes (Rettig et al., 1994 ; Scott et al., 1994b), a third has been cloned (Kvβ3), as well as splice variants of Kvβ1 (1.1, 1.2, and 1.3) (England et al., 1995 ; Heinemann et al., 1995 ; Majumder et al., 1995).

Initial expression studies focused on homotetrameric channels created by each α-subunit type ; these produced K+ currents with distinct pharmacological and biophysical properties that could be modulated by the Kvβ1.1 and Kvβ2.1 subunits (for review, see Dolly and Parcej, 1996). However, such channels did not match the full range of Kv currents recorded in various neuronal preparations. The identification of heterooligomeric channels in mammalian brain (Wang et al., 1993 ; Scott et al., 1994a) and the demonstrated ability of α subunits expressed in vitro to coassemble (Isacoff et al., 1990 ; Ruppersberg et al., 1990) meant that a huge number of subtypes of native channels might exist. If there were random assembly of just the α subunits, there could be 210 possible combinations, plus those contributed by association with the different Kvβ subunits. Therefore, it is absolutely essential to resolve this question for these functionally important Kv channels in the human CNS. Studies have shown that assembly of the Kv channel subunits is controlled because only subunits within a subfamily coassemble (Shen and Pfaffinger, 1995 ; Xu et al., 1995). The distinct regional and subcellular distribution of the Kv1 subunits also strongly suggests a mechanism determining the location of channel subtypes with the desired biophysical properties (McNamara et al., 1993 ; Veh et al., 1995 ; Bekele-Arcuri et al., 1996 ; Rhodes et al., 1996). The “clustering” proteins, e.g., SAP-97, PSD-95, and chapsin-110, also provide a further level of organisation, as shown in transfected cells (Kim et al., 1995 ; Kim and Sheng, 1996 ; Laube et al., 1996) and in vivo (Hanada et al., 1997 ; Tejedor et al., 1997).

This laboratory has addressed the subunit composition of native Kv channels in bovine cerebral cortex, using preparations purified on dendrotoxin resin (Scott et al., 1994a) and detergent-solubilised synaptic membranes (Shamotienko et al., 1997). Sequential immunoprecipitations allowed the identification of the first fully defined tetramer of α subunits for any native Kv channel. It is both clinically important and feasible to exploit such established methodologies for examining the K+ channels in human tissues, because little is known about these despite their involvement in disease states (Sanguinetti and Spector, 1997). A number of neurological conditions have been associated with altered Kv channel function. Episodic ataxia and myokymia are due to point mutations in a single gene, KCNA1 (Browne et al., 1994, 1995 ; Comu et al., 1996), that give rise to altered K+ currents and result in increased neuronal excitability. Smart et al. (1998) demonstrated that an epileptic phenotype in mice was caused by knockout of the KCNA1 gene. Mutations in non-Kv genes also change the temporal and regional expression of Kv1 subunits. The mouse strains shiverer and Trembler both show altered expression of the Kv1 channel subunits, Kv1.1 and Kv1.2 (Wang et al., 1995), even though mutations are in the genes for myelin basic protein (Roach et al., 1985 ; Molineaux et al., 1986) and peripheral myelin protein (Suter et al., 1992), respectively. This implies that changes in the state of myelination disturb ion channel organisation. The human disease Charcot-Marie-Tooth type 1A has a similar mutation as the Trembler mouse (Valentijn et al., 1992). The present investigation was undertaken on human postmortem CNS tissue with the aim of defining human Kv channel subunit assemblies, an essential first step for the design and screening of drugs directed to these targets.

MATERIALS AND METHODS

Generation and purification of Kv1.X and Kvβ specific antibodies

The peptides QVSIACTEHNLKC [Kvβ1.1 subunit residues 2-13, excluding the C-terminal cysteine (Rettig et al., 1994)] and SPARLLSLRQTGSPGMIYSTRC [Kvβ2.1 subunit residues 9-28, except for a C-terminal cysteine (Scott et al., 1994b)] were synthesised by standard Fmoc (9-fluorenylmethoxycarbonyl) solid-phase methodology (Applied Biosystems Inc.). Both peptides were coupled to bovine serum albumin via their C-terminal cysteine, as described previously (Foran et al., 1995). Immunisation of guinea-pigs with the above-noted β-subunit peptide conjugates was carried out as described (Harlow and Lane, 1988). Specific IgGs were purified by affinity chromatography on immobilised immunogen, as described previously (Shamotienko et al., 1997).

Other antibodies used were generated in rabbits and purified, as detailed elsewhere (Shamotienko et al., 1997). In brief, the peptides CNSHMPYGYAAQARARERERLAHSR [Kv1.4 α-subunit residues 13-37 (Stuhmer et al., 1989)] and CGMNHSAFPQTPFKTGN [Kv1.3 α-subunit residues 486-501, excluding the N-terminal cysteine (Stuhmer et al., 1989)] were used to generated the Kv1.4 and Kv1.3 antibodies, respectively. Fusion protein Kv1.1-pGEX (containing α-subunit residues 354-495) was used to raise the Kv1.1 antibody (Scott et al., 1994a ; Veh et al., 1995 ; Shamotienko et al., 1997). The monoclonal Kv1.2 antibody was generated in mice as described (Muniz et al., 1992).

Determination of antibody specificities

The specificity and reactivity of the α-subunit antibodies have been determined (Scott et al., 1994a ; Shamotienko et al., 1997). The α-subunit antibodies were checked for cross-reactivity in immunoblots with purified recombinant Kvβ1.1 and Kvβ2.1 subunits, and by ELISA according to the protocol of Stephenson and Duggan (1991). The cDNAs encoding full-length rat Kvβ1.1 and β2.1 were subcloned into bacterial expression vector pET15b (Novagen, U.S.A.), which attached a hexahistidine tag at the C terminus. The recombinant proteins were purified on a Ni2+ resin column (Novagen, U.S.A.).

Tissues

A total of 10 postmortem samples from seven cases without neurological disease were obtained from the NeuroResource tissue bank of the Institute of Neurology, London. Four samples of cerebral white matter, three of cerebral cortical grey matter, and three of spinal cord were examined in this study ; all were macroscopically and microscopically normal in appearance. The cases had an average age of 58 years (range 40-79 years) and an interval between death and snap-freezing of 29 h (range 11-48 h). Tissue blocks were dissected and snap-frozen on cork discs in isopentane cooled on liquid N2 and stored at -70°C for <2 years. All samples were dissected and axolemma isolated immediately following the thawing of tissue blocks.

Preparation of axolemma membranes

The procedure of Zetusky et al. (1979) was used. In brief, dissected spinal cord or white matter from human brain was finely chopped and homogenised in 0.85 M sucrose, 0.15 M NaCl, 0.01 M N-tris(hydroxymethyl)methyl-2-aminoethanesulphonic acid (TES), pH 7.5, and 0.3 mM phenylmethylsulphonyl fluoride (PMSF). Following centrifugation at 82,000 g for 15 min, the floating layer of myelinated axons was resuspended in the above buffer and recentrifuged. These were osmotically shocked in 0.01 M TES, pH 7.5, pelleted, and resuspended in 0.65 M sucrose. The membranes were isolated by discontinuous sucrose gradients. The axolemma-enriched fractions were harvested from the 0.8/1.0 M and the 1.0/1.2 M sucrose interfaces, combined, diluted two-to fourfold with 10 mM TES, pH 7.5, and pelleted.

Isolation of synaptic membranes

Synaptic plasma membranes were prepared from human cerebral cortical grey matter as described elsewhere (Bennett et al., 1986 ; Parcej and Dolly, 1989). They were not harvested from spinal cord due to the limited supply of tissue and the focus upon the axonal membranes, which yielded a cleaner preparation. In brief, dissected grey matter was homogenised in 0.32 M sucrose containing 7 mM imidazole-HCl, pH 7.5, 2 mM EDTA, 25 μg/ml bacitracin, 10 μg/ml soya bean trypsin inhibitor, 0.2 mM benzamidine, and 0.1 mM PMSF. The homogenate was centrifuged at 800 g for 10 min ; the pellet was resuspended in the above buffer and recentrifuged. The supernatants were pooled and centrifuged at 34,000 g for 45 min. The resulting pellet was resuspended in 1 M sucrose, 7 mM imidazole-HCl, pH 7.5, 2 mM EDTA, 25 μg/ml bacitracin, 10 μg/ml soya bean trypsin inhibitor, 0.2 mM benzamidine, and 0.1 mM PMSF and centrifuged at 34,000 g for 60 min. The pellet was washed and resuspended in the above 0.32 M sucrose buffer solution.

Solubilisation of membranes

Membranes were solubilised at 1-2 mg of protein/ml, as previously described (Parcej and Dolly, 1989) with some modifications. Aliquots of axolemma or synaptic membranes were incubated 1:1 (vol/vol) in extraction buffer [0.125 M imidazole-HCl, pH 7.5, 0.5 M KCl, 5 mM EDTA, 0.3 mM PMSF, 1 mM benzamidine, 0.5 mg/ml bacitracin, 0.5 mg/ml soya bean trypsin inhibitor, and 4% (wt/vol) Thesit]. Following extraction, the mixture was diluted twofold with 0.125 M imidazole-HCl, pH 7.5, and 0.5 M KCl and centrifuged for 20 min at 30,000 g and for 45 min at 100,000 g at 4°C.

Immunoprecipitation and immunoblotting

These procedures were carried out as described by Shamotienko et al. (1997) with the following modifications. Aliquots of the solubilised membrane (0.5-1 ml) were incubated for 5-8 h at 4°C with 20 μg/ml of each β-subunit-specific antibody and 50-70 μl of anti-rabbit or anti-mouse IgG-agarose concurrently. Following centrifugation, the supernatants were removed for further rounds of immunoprecipitation and the pellets washed four times with 20 mM Tris-HCl, pH 7.4, 150 mM NaCl containing 1% (wt/vol) Thesit, then once in buffer without Thesit, and resuspended in sample buffer. Synaptic membranes and immunoprecipitated samples were subjected to sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) on 8 or 10% gels for α or β subunits, respectively. Separated proteins were transferred to polyvinylidene difluoride membrane, as detailed by Scott et al. (1994a). The membranes were then blocked with 5% (wt/vol) dried milk and 0.5% (vol/vol) Tween 20 in 20 mM Tris-HCl, pH 7.4, 150 mM NaCl, and probed overnight at 4°C with 1-3 μg/ml rabbit or mouse purified anti-Kv1.X or guinea-pig anti-Kvβ subunit IgGs. Immunoreactivity was detected using the enhanced chemiluminescence system (Amersham) in conjunction with horseradish peroxidase-conjugated goat anti-rabbit, anti-mouse, or anti-guinea-pig IgG.

RESULTS

Initial analysis of the Kv1.X subunit contents of white and grey cerebral matter from human brain and of spinal cord

The specificities of the α-subunit antibodies used have been established by Shamotienko et al. (1997). All proved to be subunit isoform-specific except for the anti-Kv1.1 IgG, which displayed cross-reactivity with the Kv1.4 subunit ; this did not create a problem for blotting, as these two subunits are distinguishable by their different molecular weights. The α subunits Kv1.1, 1.2, and 1.4 were present in the three tissue types analysed (Fig. 1), each showing molecular sizes of 80, 78, and 96 kDa that are similar to those found for bovine brain (Scott et al., 1994a). Kv1.4 appeared to run as a doublet (Fig. 1), although this was not so apparent with the immunoprecipitated samples (see later). Immunoprecipitation with the Kv1.3 antibody (Fig. 2) was necessary to reveal the presence of Kv1.3 in brain grey matter (Fig. 3A), but this subunit was not found in white matter (Fig. 4) or spinal cord (Fig. 5). The relative amounts of each subunit varied between the tissues. Using an equivalent quantity of protein, spinal cord showed noticeably less Kv1.2 and 1.4 than either of the brain samples. The Kv1.1 subunit appeared equally abundant in brain white matter and spinal cord, with a lower content in grey matter (Fig. 1). A nonspecific band was also detected in the latter tissue, running directly above the Kv1.1 subunit ; this was not found in other tissues or in immuno-precipitated samples from grey matter (Fig. 3) and is, thus, attributed to cross-reactivity, possibly by the secondary anti-rabbit IgG, with an irrelevant protein. The distribution of Kv1.2 subunit between the grey and white matter appeared similar, whereas brain grey matter had a higher proportion of Kv1.4 than white matter. Unfortunately, the results detailed herein do not allow the determination of the rank order of abundance of the subunits in these tissues due to the differing affinities of both the primary and secondary antibodies.

Figure 1.

Identification of the Kv1.X α subunits present in membranes from human CNS. Crude membranes (40 μg of protein) from grey matter (GM), white matter (WM), and spinal cord (SpC) were subjected to SDS-PAGE, transferred to polyvinylidene difluoride membrane, and probed with 1-3 μg/ml anti-Kv1.X IgG, as indicated. Labelled blots were visualised using secondary antibody conjugated to horseradish peroxidase, in conjunction with the enhanced chemiluminescence system. The synaptic membrane GM samples probed with rabbit polyclonal anti Kv1.X IgGs show artifactual bands, which may be due to a nonspecific interaction with the secondary antibody. Positions of protein standards are shown in kDa.

Figure 2.

General diagram of the standard multiple-step protocol for sequential immunoprecipitation of Kv1 channels from detergent-solubilised neuronal membranes. The Kv1.X antibodies were used initially in the order shown : each pellet was probed with all of the antibodies detailed in Fig. 1. Further data were then obtained by modifying the order in which the antibodies were used.

Figure 3.

Immunoprecipitation of Kv1 channels from solubilised synaptic membranes prepared from brain grey matter. One milliliter of detergent-solubilised membranes was immunoprecipitated with the anti-Kv1.X antibodies in the sequences shown. The resultant pellets were subjected to SDS-PAGE and blotted as described in Fig. 1. A-D : The pellets were probed with the α subunit, anti-Kv1.X antibodies specified on the left of the blots. E : The pellets were probed with the anti-Kvβ IgGs indicated to the left. Samples of recombinant Kvβ1.1 or Kvβ2.1 were also run as positive controls.

Figure 4.

Immunoprecipitation of Kv1 channels from solubilised axolemma membranes prepared from brain white matter. A-C : An aliquot (1 ml) of detergent-solubilised membranes was immunoprecipitated with the anti-Kv1.X antibodies in the order indicated and immunoblotting performed as described in Fig. 3. D : The immunoprecipitations were carried out as above and blotted with the anti-Kvβ IgGs indicated ; samples of recombinant Kvβ1.1 or Kvβ2.1 were also run as described in Fig. 3.

Figure 5.

Sequential immunoprecipitation of Kv1 channels from solubilised axolemma membranes prepared from spinal cord tissue. A-C : An aliquot (1 ml) of detergent-solubilised membranes was immunoprecipitated with the anti-Kv1.X antibodies in the standard, alternate, and reverse sequences shown, and blotted as in Fig. 3. D : The immunoprecipitations were carried out as above and blotted with the anti-Kvβ IgGs indicated ; samples of recombinant Kvβ1.1 or Kvβ2.1 were also run as described in Fig. 3.

FIG. 1.

FIG. 2.

FIG. 3.

FIG. 4.

FIG. 5.

Determination of subunit combinations

The Kv1.1, 1.2, 1.4, and 1.3 antibodies were used for the immunoprecipitation studies. In view of the minor cross-reactivity of anti-Kv1.1 IgG with Kv1.4, it was preferable to use Kv1.4 antibody before the IgG selective for Kv1.1. However, the limited cross-reactivity of anti-Kv1.1 IgG resulted in only a minimal depletion of the pool of Kv1.4 subunit when used initially. Figure 2 shows a schematic of the sequential immunoprecipitation protocol. Depending on the results obtained from the initial precipitations for each tissue, further sequential immunoprecipitations were carried out using alternate orders of the Kv1.X antibodies, although the general methodology remained the same. Initial studies, similar to those described by Shamotienko et al. (1997), were carried out to ensure that sufficient quantities of primary and secondary antibodies were added to the solubilised membranes to ensure complete precipitation of the desired subunit (data not shown).

Cerebral cortical grey matter.

Figure 3A shows the results of sequential immunoprecipitation from grey matter, using anti-Kv1.X antibodies in the order of those specific for the least to the most abundant subunit, as has been established for bovine brain (Scott et al., 1994a ; Shamotienko et al., 1997). The Kv1.3 antibody successfully precipitated Kv1.4, 1.1, and 1.2 subunits. The pair combinations Kv1.3/1.4, 1.3/1.1, and 1.3/1.2 could explain this. However, the low abundance of the Kv1.3 subunit (Shamotienko et al., 1997), as compared with the other subunits, could indicate a potential tetramer of Kv1.3/1.4/1.1/1.2 subunits, although the reverse immunoprecipitation did not allow the reliable detection of the Kv1.3 subunit (Fig. 3B). This could be due to the inability of the Kv1.3 antibody to detect the low level of its antigen present. However, previous studies on synaptic membranes from bovine cerebral cortex have shown that all of the Kv1.3 subunit is coassociated with Kv1.4 or 1.2 proteins (Shamotienko et al., 1997). Precipitation of the Kv1.3-depleted supernatant with the anti-Kv1.4 antibody brought down the remainder of the Kv1.4 subunit together with Kv1.1 and 1.2 (Fig. 3A). Moreover, the relative intensities of the Kv1.4 bands in Fig. 3A show that the majority of the Kv1.4 subunit was not coassociated with the Kv1.3 subunit, but coassembled with Kv1.1 and 1.2. As these subunits could be associated in a single oligomer or in pairs, i.e., Kv1.4/1.1 and Kv1.4/1.2, alternate series of sequential immunoprecipitations were carried out to distinguish these possibilities (Fig. 3B and C). An initial precipitation using the anti-Kv1.2 IgG depleted the solubilised membranes of all the Kv1.2-containing channels ; Kv1.1 and 1.4 were also found in the pellet (Fig. 3B). Nevertheless, a second precipitation with Kv1.1 antibody allowed the detection of both Kv1.1 and 1.4 in the resultant pellet, but not Kv1.2. Unfortunately, the tendency of the anti-Kv1.1 antibody to cross-react with the Kv1.4 subunit complicates the interpretation of this result. Nevertheless, it demonstrated that not all of Kv1.1 or Kv1.4 is coassociated with Kv1.2 ; a semiquantitative comparison of the Kv1.1 bands in Fig. 3B suggests that ~60% of the Kv1.1 subunit was associated with the Kv1.2 subunit. A further round of immunoprecipitation using the anti-Kv1.4 IgG yielded Kv1.4 alone, with Kv1.1, 1.2, and 1.3 being absent from the pellet. This suggests a Kv1.4 homooligomer ; again, the intensity and size of the bands show that a substantial proportion of the Kv1.4 is not associated with either Kv1.2 or Kv1.1 (Fig. 3B). The presence of the Kv1.4/1.2 combination is demonstrated in Fig. 3C, where an initial removal of all the Kv1.1 oligomer still allowed Kv1.4 to be coprecipitated by anti-Kv1.2 IgG. The final precipitation of the series (Fig. 3C) again revealed the abundant Kv1.4 homooligomer. The penultimate precipitation with anti-Kv1.1 IgG of the initial series (Fig. 3A) sedimented the remainder of the Kv1.1 subunit and Kv1.2, demonstrating a Kv1.1/1.2 combination. The following and final immunoprecipitation revealed a Kv1.2 homooligomer ; using a supernatant devoid of Kv1.3, 1.4, and 1.1, precipitation with the monoclonal anti-Kv1.2 antibody brought down yet more Kv1.2 subunit (Fig. 3A). As homooligomers of both Kv1.2 and 1.4 were shown (Fig. 3A-C), the presence of a Kv1.1 homooligomer was investigated (Fig. 3D). The initial two precipitations in the series brought down all the Kv1.4 and 1.2 subunits and coprecipitated Kv1.1. The final immunoprecipitation with anti-Kv1.1 IgG did not bring down any more Kv1.1 subunit (Fig. 3D), suggesting that no Kv1.1 homooligomer is present. No further novel information was obtained from this sequential precipitation series (Fig. 3D) ; however, the coassociations identified confirmed the previous data.

Parcej et al. (1992) showed that the majority of the Kv1.X channels are octomers of four α and four β subunits. Therefore, following the establishment of the α-subunit combinations, the presence of the auxiliary subunits Kvβ1.1 and Kvβ2.1 was investigated (Fig. 3E). No Kvβ1.1 was detectable in any of the immunoprecipitates, although recombinant Kvβ1.1 could be visualised readily (Fig. 3E). Neither was Kvβ2.1 coprecipitated by the Kv1.3 or 1.4 antibodies. However, both the third and fourth immunoprecipitations in the series with anti-Kv1.1 and anti-Kv1.2 IgGs, respectively, successfully coprecipitated Kvβ2.1 (Fig. 3E) ; a lower band was also seen in the anti-Kv1.1 precipitate, which may be due to proteolysis, as it was not detected in any other immunoprecipitate. This result established that the Kv1.2 homomer is associated with Kvβ2.1 in cerebral grey matter ; however, it cannot be deduced that a Kv1.1/β2.1 combination is also present, because the association of Kv1.1 with 1.2 was already shown (Fig. 3A). To determine if the Kvβ2.1 subunit was associated with Kv1.1 and Kv1.2 subunits individually, the reverse order of precipitation was also examined. It was found that an initial precipitation with anti-Kv1.2 IgG coprecipitated all the Kvβ2.1 subunit (Fig. 3C). The following immunoprecipitation with Kv1.1 antibody did not bring down any more Kvβ2.1. Thus, all the Kvβ2.1 was coassociated with the Kv1.2 subunit (Fig. 3E), implying that the heterooligomers Kv1.1/1.2/β2.1 and Kv1.2/β2.1 exist in cerebral grey matter.

Cerebral white matter.

Initially, the antibodies were used for precipitation (Fig. 4A) in the previously established order of those reactive with the least to the most abundant subunit (Scott et al., 1994a ; Shamotienko et al., 1997). No Kv1.3 subunit could be detected in the initial pellet ; likewise, the anti-Kv1.3 IgG failed to sediment any other Kv1.X subunits (Fig. 4A). However, the second round of precipitation with the anti-Kv1.4 antibody pelleted both Kv1.4 and 1.1, but no Kv1.2, implying the coassociation of Kv1.4 with 1.1 in the absence of Kv1.2. The third precipitation with anti-Kv1.1 IgG precipitated more Kv1.1 together with a fraction of the Kv1.2 subunit ; a final precipitation with the anti-Kv1.2 IgG removed the rest of the Kv1.2 subunit (Fig. 4A). Thus, a Kv1.1/1.2 combination does exist in white matter, as well as an apparent Kv1.2 homooligomer. To confirm the absence of the Kv1.4/1.2-containing combination, a reverse immunoprecipitation was carried out ; using anti-Kv1.2 IgG for the initial precipitation, Kv1.2 and 1.1 were found in the pellet, but not Kv1.4 (Fig. 4B). The Kv1.1/1.4 oligomer was then removed using anti-Kv1.1 IgG (not shown) and the resultant supernatant was subjected to precipitation with the anti-Kv1.4 antibody ; as expected, this allowed the detection of Kv1.4, but not Kv1.2 (Fig. 4B). As for cerebral grey matter (Fig. 3D), the presence or not of a Kv1.1 homooligomer was looked at by an alternate order of immunoprecipitation (Fig. 4C). The solubilised membranes were subjected initially to anti-Kv1.4 IgG, followed by anti-Kv1.2 IgG ; these brought down Kv1.4 and 1.1, 1.2 and 1.1 subunits, as appropriate (Fig. 4C). A final immunoprecipitation with anti-Kv1.1 IgG did not sediment any further subunits (Fig. 4C), confirming the absence of a Kv1.1 homooligomer in this tissue.

The presence of the Kvβ subunits was also examined ; Kvβ1.1 and Kvβ2.1 were both detected (Fig. 4D), unlike cerebral grey matter (Fig. 3E). The initial precipitation

with Kv1.4 antibody brought down a very minor amount of the Kvβ1.1, but no Kvβ2.1 (Fig. 4D). A second round of immunoprecipitation with anti-Kv1.1 IgG gave a greater amount of Kvβ1.1, but still no Kvβ2.1. The final immunoprecipitation with the Kv1.2 specific antibody coprecipitated Kvβ2.1 and the major proportion of Kvβ1.1 (Fig. 4D). The reverse order of precipitation, using anti-Kv1.2 IgG initially, allowed the detection of Kvβ1.1 with Kv1.2 (Fig. 4D), confirming that the major pool of this β subunit is associated with the Kv1.2 and/or the Kvβ2.1 subunit. This latter subunit was only coassociated with Kv1.2, as shown by both sets of precipitations (Fig. 4D). The absence of Kvβ2.1 association with either Kv1.1 or 1.4 implies that a substantial proportion of the Kv1 channels in the axolemma membranes from brain white matter are devoid of this subunit, although some of these channels, Kv1.4/1.1 and Kv1.1/1.2, are associated with the Kvβ1.1 subunit. The association of auxiliary subunits with the Kv1.2 homooligomer is more complicated, with a number of possible combinations that cannot be distinguished in this study : Kv1.2 and β1.1, Kv1.2 and β2.1, or Kv1.2, β1.1, and β2.1. The recombinant Kvβ1.1 and β2.1 occasionally gave a doublet with a weaker upper band ; this may be due to incorrect folding of the recombinant protein.

Spinal cord.

As in the case of brain white matter, Kv1.3 subunit could not be detected in spinal cord axolemma membranes and no subunits were precipitated by the anti-Kv1.3 IgG (Fig. 5A). Continuing the sequential immunoprecipitations in the order illustrated in Fig. 2 revealed the combinations Kv1.4/1.1 and Kv1.1/1.2, but not Kv1.4/1.2 (Fig. 5A), a pattern similar to that seen in brain white matter (Fig. 4A). However, all the Kv1.2 subunit was precipitated with the anti-Kv1.1 IgG (Fig. 5A), excluding the possibility of any Kv1.2 homooligomer. To confirm these results, the sequential immunoprecipitations were carried out using alternate orders of the anti-Kv1.X IgGs ; this would also allow the detection of a possible Kv1.1 or Kv1.4 homooligomer. An initial precipitation with the anti-Kv1.1 antibody removed all of the Kv1.1 subunit and coprecipitated Kv1.2 and 1.4 (Fig. 5B) ; subsequent precipitation of the supernatant with the anti-Kv1.4 revealed only Kv1.4 in the pellet (Fig. 5B). Further probing of the Kv1.1- and Kv1.4-depleted supernatant with the anti-Kv1.2 specific antibody demonstrated that there was no Kv1.2 subunit remaining (Fig. 5B), reaffirming that all the Kv1.2 subunit is associated with Kv1.1. To investigate if there was a potential Kv1.1 homooligomer, an aliquot of the membrane extract was reacted initially with the anti-Kv1.2 antibody ; as expected, this precipitated Kv1.2 and 1.1, but no Kv1.4 (Fig. 5C), thereby reaffirming the coassociation of Kv1.1 with Kv1.2. A subsequent precipitation with anti-Kv1.4 IgG pelleted Kv1.4 and more Kv1.1, confirming the coassembly of Kv1.4 and 1.1. However, a third precipitation of the Kv1.2- and 1.4-depleted supernatant did not detect any Kv1.1 subunit (Fig. 5C), demonstrating that all of the Kv1.1 subunit is coassociated with either Kv1.2 or Kv1.4 ; thus, there is no Kv1.1 homooligomer. These findings unveil the presence of at least three subtypes of Kv1.X channels in spinal cord, i.e., Kv1.1/1.2, Kv1.1/1.4, and Kv1.4 ; the presence of other unidentified subunits has not been excluded. Probing of α-subunit immunoprecipitates did not allow the detection of any of the Kvβ subunits, unlike the previous tissues (Fig. 5D) ; the antibodies did detect the recombinant β subunits. A doublet was seen for the recombinant Kvβ1.1 protein ; again, this is possibly due to incorrect folding of the subunit.

DISCUSSION

It has been established that the Kv1 channel α and β subunits coassociate in vivo and in vitro (Isacoff et al., 1990 ; Ruppersberg et al., 1990 ; Parcej et al., 1992 ; Sheng et al., 1993 ; Wang et al., 1993 ; Rettig et al., 1994 ; Scott et al., 1994a,b). Also, in situ hybridisation and immunocytochemical experiments on rodents have localised the individual Kv1.X subunits (Beckh and Pongs, 1990 ; Sheng et al., 1993 ; Wang et al., 1993 ; Veh et al., 1995). However, unlike this investigation, the latter studies did not allow determination of the oligomeric composition of these K+ channels. A previous study from this laboratory on bovine cortex established the first fully defined Kv1 tetramer (Shamotienko et al., 1997) ; a similar examination of the Kv1 channels in human CNS was accomplished herein, for the first time. An initial investigation of the individual subunits present was followed by immunoprecipitations to decipher the actual combinations in human brain white and grey matter, as well as spinal cord. It is important to stress that the conditions of the immunoprecipitations were such that each subunit was precipitated quantitatively by its specific antibody.

The first stage of the work clearly showed the presence of the α subunits Kv1.1, 1.2, and 1.4 in all the tissues studied ; variations in abundance and proportions of the subunits in each tissue suggested differences in their K+-channel compositions. The Kv1.3 subunit could not be detected by direct immunoblotting, but it was visualised following enrichment by immunoprecipitation in cerebral grey matter only ; this low abundance and restricted distribution of Kv1.3 accord with the findings for bovine cortex (Shamotienko et al., 1997). The immunoprecipitation protocols revealed a limited number of subunit combinations (Table 1), thereby establishing that only a small proportion of the possible oligomers occurred in the three tissues, although these varied in number and complexity. Synaptic membranes from brain grey matter displayed the greatest variety, possessing seven distinct α-subunit oligomers. A potential tetramer (Kv1.3/1.4/1.1/1.2) was identified exclusively in the grey matter. This proposed combination is supported by the fact that anti-Kv1.3 IgG coprecipitated the Kv1.1 subunit, which always coexists with Kv1.2 or 1.4. Thus, the triple combinations Kv1.3/1.1/1.4 and Kv1.3/1.1/1.2 must be present. Moreover, the Kv1.4/1.2 combination is also found within the cortical grey matter. Thus, although the tetramer cannot be proven absolutely, its presence seems highly probable, especially given the low abundance of the Kv1.3 subunit. The proposed tetramer was not present in white matter or spinal cord, Kv1.3 being undetectable, nor was the Kv1.4/1.1/1.2 combination seen in these tissues (Table 1). The combinations where one to three different subunits were identified are likely to contain a further copy of one of these subunits (the immunoprecipitation technique does not allow multiple copies of the same subunit within a given channel complex to be distinguished) ; the other possibility is the presence of an unknown subunit. A further major distinction between the synaptic preparation from grey tissue and the axolemma membranes from white tissue and spinal cord was the complete lack of association of Kv1.2 and 1.4 in the latter samples (Table 1), precluding their simultaneous presence in any tetramers occurring in these tissues. The axolemma membranes from white matter and spinal cord displayed a similar pattern of α-subunit combinations, distinguished only by the presence of the Kv1.2 homooligomer in the former. This homooligomer was also seen in the brain grey matter, along with the ubiquitous Kv1.4 homooligomer (Table 1). However, no tissue contained a Kv1.1 homooligomer.

Table 1. Kv1 α- and β-subunit combinations in human CNSSequential immunoprecipitations established the α and β combinations described above. Where four distinct α subunits were not found, a further copy of an identified subunit or possibly one for which antibodies were not available is the only other candidate. The Kv1.X subfamily has been shown to be an octomeric protein complex containing four α and four β subunits (Parcej et al., 1992). Therefore, the demonstrated α/β-subunit associations are also shown.
Human tissueDemonstrated Kv1.X subunit combinations
Cerebral cortical grey matter1.3 + 1.4 + 1.1 + 1.2
 1.4 + 1.1 + 1.2
 1.4 + 1.1
 1.4 + 1.2
 1.1 + 1.2
 1.4
 1.2
 1.1 + 1.2 + β2.1
 1.2 + β2.1
Cerebral white matter1.4 + 1.1
 1.1 + 1.2
 1.4
 1.2
 1.4 + 1.1 + β1.1
 1.1 + 1.2 + β1.1
 1.2 + β1.1
 1.2 + β2.1
Spinal cord1.4 + 1.1
 1.1 + 1.2
 1.4

TABLE 1.

Earlier studies had established a rank order of Kv1.X abundance in synaptic membranes from bovine cortex, with Kv1.2 being the most abundant α subunit and most oligomers containing at least one copy (Scott et al., 1994a ; Shamotienko et al., 1997). This differs from the results described herein for human tissue where some of the identified multimers did not contain any Kv1.2 (Table 1) ; although this could be due to a species difference, it is more characteristic of the combinations found in white matter and spinal cord. However, all the tissues contained the Kv1.1/1.4 heterooligomer, in agreement with the colocalisation of Kv1.1 and 1.4 in the absence of Kv1.2, as observed in rat tissue by immunohistochemical staining (Rhodes et al., 1997).

A second prominent combination, Kv1.1/1.2, was also found in all the human tissues. Accordingly, it has been identified in bovine cortex and rat cerebellum (Koch et al., 1997 ; Shamotienko et al., 1997), and this pair of subunits was shown to be colocalised in rat brain and spinal cord (Veh et al., 1995). The prevalence of Kv1.1 is of great interest because many disease states, due to mutations in Kv1 genes or causing alterations in the function or localisation of K+ channels, involve this subunit (Browne et al., 1994, 1995 ; Wang et al., 1995 ; Comu et al., 1996 ; Smart et al., 1998). The proposed channel tetramer, Kv1.3/1.4/1.1/1.2, in cortical grey matter is distinct from that previously identified in bovine cortex, Kv1.3/1.4/1.2/1.6 (Shamotienko et al., 1997), due to the association of Kv1.3 and 1.1 and the absence of Kv1.6. These tetramers may be expected to show distinct electrophysiological profiles due to absence or presence of the Kv1.6 subunit ; the latter has a domain (NIP) that can prevent N-type fast inactivation of K+ channels by Kv1.4 or β1.1 subunits (Roeper et al., 1998).

Parcej et al. (1992) established that the majority of Kv1 channels in bovine cortex contain auxiliary Kvβ subunits ; in fact, all the Kv channel oligomers identified in bovine synaptic membranes possess the Kvβ2.1 subunit (Shamotienko et al., 1997). This was not the case in the present study ; only two of the identified combinations in cerebral cortical grey matter contained the Kvβ2.1 subunit : Kv1.1/1.2 and the Kv1.2 homooligomer (Table 1). Furthermore, an initial precipitation with anti-Kv1.2 IgG demonstrated that all of Kvβ2.1 was associated with the Kv1.2 subunit. This was also the situation in cerebral white matter, where Kvβ2.1 could be coprecipitated by Kv1.2 antibodies only. However, unlike grey matter, Kvβ1.1 was detected in cerebral white matter. The coassociation of the Kvβ1.1 subunit was more widespread than Kvβ2.1, with it being coimmunoprecipitated by anti-Kv1.4, 1.1, and 1.2 IgGs. Both Kvβ1.1 and β2.1 were coprecipitated by anti-Kv1.2 IgG, suggesting that they might coexist in a single oligomer. This is supported by Rhodes et al. (1996), who showed that Kvβ2.1 is the predominant Kvβ subunit in rat brain and Kvβ1.1 coexists with it in complexes. However, Kvβ1.1-containing oligomers, without Kvβ2.1, are clearly present (Table 1). The spinal cord membranes were devoid of all the auxiliary subunits. However, the possibility exists that they are present at levels below detection. The expression of Kvβ3 mRNA is prominent only in the olfactory bulb and thalamic nuclei (Heinemann et al., 1995) ; thus, it is unlikely to be found in these samples. The auxiliary subunits have distinct functions : Kvβ1 confers rapid inactivation on noninactivating α subunits (Rettig et al., 1994 ; Heinemann et al., 1996), and Kvβ2.1 behaves as a chaperone for Kv1.2 (Rhodes et al., 1996) and modifies the gating of Kv1.4 (McCormack et al., 1995 ; McIntosh et al., 1997). It also affects the Kvβ1.1-mediated K+-current inactivation (Xu and Li, 1997). The absence of detectable Kvβ1.1 subunits in cerebral grey matter and spinal cord implies that only Kv1.4-containing combinations can inactivate rapidly and generate fast transient A-type currents, and these channels also lack the Kvβ2.1 subunit (Table 1). The detection of Kvβ1.1 in cerebral white matter suggests that there may be a greater number of rapidly inactivating channels in these axolemma membranes, as compared with the other CNS areas studied. A-type currents have been identified previously in axons of neuronal cells (Sheng et al., 1992 ; Debanne et al., 1997 ; Cooper et al., 1998), but were assigned generally to the presence of Kv1.4 subunit, although the Kvβ1.1 subunit was not considered. The individual biophysical characteristics of fast transient K+ currents due to the presence of Kvβ1.1 could be subtly distinct, as the number of associated Kvβ1.1 subunits can vary from one to four (Xu et al., 1998). Moreover, some may also be associated with Kvβ2.1.

The presence of distinct α-subunit combinations identified in the separate tissues and the differences in α/β-subunit associations strongly support the idea that each possesses a wide range of Kv channels with distinct pharmacological and biophysical characteristics. This study also implies that the Kv1 subunit assembly is not random ; instead, it would appear that the properties of each channel type are tailored for a particular role. Examination of these heterooligomers by recombinant methodologies, coupled with biochemical and electrophysiological techniques, should yield insights into their respective functions. The data presented herein are the initial step in the analysis of Kv1 channels in disease states ; changes in subunit combination or localisation and, by implication, their electrophysiological properties need to be examined. Comparison of normal and pathological channel phenotypes would provide vital information on their respective functions in the nervous system.

Ancillary