INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Potassium channels constitute a very diverse group of ion channels and are composed of variable combinations of subunits encoded in large multigene families (Chandy and Gutman 1995; Coetzee et al. 1999; Hille 1992; Jan and Jan 1990; Pongs 1992; Rudy 1988). This diversity contributes to the ability of specific neurons to respond uniquely to different inputs. One of these groups of genes, the Kv family, encodes subunits of tetrameric voltage-gated K⁺ channels and is divided into several subfamilies based on sequence similarities and hence probable evolutionary relationships (Chandy and Gutman 1995; Coetzee et al. 1999; Jan and Jan 1990; Pongs 1992). Subunits of the same subfamily, but not from different subfamilies, can form heteromeric channels, suggesting that each group of genes encodes a distinct system of channels (reviewed in Coetzee et al. 1999).

One of the Kv subfamilies known as Kv3 has generated particular interest because of the special roles of K⁺ channels containing Kv3 subunits in enabling high-frequency repetitive firing (Erisir et al. 1999; Wang et al. 1998; reviewed in Rudy and McBain 2001; Rudy et al. 1999). The Kv3 subfamily consists of four genes (Kv3.1, Kv3.2, Kv3.3, and Kv3.4). Each Kv3 gene encodes multiple products by alternative splicing of 3' ends resulting in the expression of K⁺ channel proteins that differ only in short C-terminal sequences (Luneau et al. 1991a,b; Rudy et al. 1992, 1999; Vega-Saenz de Miera et al. 1992).

A functional role for the alternative splicing of Kv3 genes has not been established because Kv3 alternatively spliced products from the same gene express virtually identical currents in heterologous expression systems (Vega-Saenz de Miera et al. 1994). However, it has been proposed that the alternative splicing might be implicated in generating isoforms mediating differential subcellular targeting and modulation (McIntosh et al. 1998; Vega-Saenz de Miera et al. 1994). In support of a role for the differential alternative splicing in membrane targeting, Ponce et al. (1997) observed that different Kv3 splice versions were targeted to different membrane domains in polarized epithelial Madin-Darby canine kidney (MDCK) cells. However, the subcellular distribution of different Kv3 splice versions in neurons has not been studied.

To test the hypothesis that the alternative-splicing of Kv3 genes plays a role in determining the subcellular localization of native Kv3 channels in neurons, we raised antibodies specific for the two known spliced versions of the Kv3.1 gene, Kv3.1a and Kv3.1b, and used these antibodies to investigate their cellular and subcellular localizations in mouse brain.

We present evidence that Kv3.1a and Kv3.1b proteins have different subcellular patterns of expression in CNS neurons. These results lend support to the hypothesis that the alternative splicing of Kv3 genes is involved in regulating the channel's subcellular localization. This suggests that the variant C termini contain or regulate targeting signals and raises the possibility that the alternative splicing of 3' ends of other K⁺ channel genes (Isomoto et al. 1996; Lesage et al. 1995; Pan et al. 2001; Tian et al. 2001) has similar roles. Some of the results shown here have been presented in abstract from (Ozaita et al. 1998, 2000).

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Antibody production

The peptide Ac-CPLAQEEILEINRAGRKPLRG-NH₂ was synthesized by Quality Controlled Biochemicals (QCB; Hopkinton, MA). This peptide, corresponding to the carboxyl terminal sequence of the Kv3.1a protein (residues 488-508; see diagram in Fig. 1A), was coupled via the cysteine to keyhole limpet hemocyanin (KLH). The KLH-linked Kv3.1a peptide was injected into rabbits using standard procedures for antisera production by QCB. To purify Kv3.1a-specific antibodies from the antisera (see RESULTS), a new peptide (Ac-C[AHX]NRAGRKPLRG-amide) containing the alternatively spliced C terminus (residues 499-508) of Kv3.1a, was coupled via the cysteine to Sulfolink-Sepharose resin (Pierce, Rockford, IL) and antibodies purified by affinity chromatography following the supplier's protocol. The preparation and characterization of Kv3.1b-specific antibodies have been previously described (Weiser et al. 1995).

View larger version (40K): [in this window] [in a new window]   Fig. 1. Characterization of Kv3.1a- and Kv3.1b-specific antibodies. A: sequence of the C-terminal area of the Kv3.1a (top) and Kv3.1b (bottom) proteins near the splice junction (arrow). The complete alternative C-terminus of Kv3.1a is shown, but the Kv3.1b variant has 74 additional residues that are not shown in the diagram. A primary peptide (bold sequence in Kv3.1a) coupled to KLH was used for immunization, and a secondary peptide (underlined sequence) coupled to aminohexanoic acid linkers was used to affinity purify specific antibodies against the Kv3.1a variant. B: immunoblots of Kv3.1a and Kv3.1b proteins heterologously expressed in HEK-293T cells. Extracts from cells transfected with Kv3.1a or Kv3.1b cDNAs were size fractionated by 10% SDS-PAGE. Samples were transferred to nitrocellulose membranes and probed with purified Kv3.1a antibody (left) or Kv3.1b antibody (right). Bound antibody was detected by ECL autoradiography. Numbers in between the panels denote molecular weights (MW) in kDa obtained from prestained MW standards. C: immunoblots of mouse brain membrane extracts from wild-type (+/+) and Kv3.1 knockout (/) mice. Nitrocellulose membranes were probed with purified Kv3.1a antibody (left) or Kv3.1b antibody (right). Numbers between panels denote MW in kDa obtained as in B. D: immunofluorescence staining of Kv3.1a and Kv3.1b subunits expressed in HEK-293T cells. HEK-293T cells were transfected with Kv3.1a (Kv3.1a cells, left) or with Kv3.1b (Kv3.1b cells, right) cDNAs. Transfected cells were fixed, permeabilized, and incubated with purified Kv3.1a antibody (top) or with Kv3.1b antibody (bottom). E: Kv3.1a and Kv3.1b protein expression in the cerebellum. Consecutive coronal sections obtained from a wildtype mouse (Kv3.1 +/+, left) or a Kv3.1 knockout mouse (Kv3.1 /, right) were probed with Kv3.1a antibody (top) or with Kv3.1b antibody (bottom). Sections were stained with the DAB reaction (see METHODS). Scale Bar in E: 6 µm for D and 1.5 mm for E.

Transient transfection of HEK-293T cells

Transfected and untransfected HEK-293T (a T-antigen-expressing derivative of HEK-293, human embryonic kidney) cells were cultured in DMEM, pH 7.4 (Life Technologies, Rockville, MD), supplemented with 10% fetal bovine serum (Life Technologies) in the presence of penicillin and streptomycin at 37°C in a 95% O₂-5% CO₂ atmosphere in 60-mm-diam culture dishes (Becton Dickinson, Franklin Lakes, NJ). HEK-293T cells were transiently transfected after reaching 40% confluence with Kv3.1a. or Kv3.1b (Luneau et al. 1991b) cDNAs in pcDNA3 (Invitrogen, Carlsbad, CA) using LipofectAMINE reagent (Life Technologies) following the manufacturer's protocol. Typical transfections yielded ~50% efficiency. Forty-eight hours after transfection the cells were washed twice with cold PBS and incubated on ice with 1 ml cold TNEE [(in mM) 50 Tris, 150 NaCl, 1 EDTA, and 1 EGTA, pH 7.4] plus 1% Triton X-100 for 30 min. Cells were then scraped off and centrifuged in a microcentrifuge at 4°C for 30 min to remove non-solubilized material. The top two-thirds of the supernatant were collected for further experiments. Protein concentration was measured with a colorimetric protein assay (Bio-Rad, Hercules, CA). A protease inhibitor mixture was added [0.5% (vol/vol); Sigma, St. Louis, MO].

Mouse brain membrane extracts

Mouse brain membrane extracts were prepared from a P3 fraction of tissue homogenate (Hartshorne and Catterall 1984) and solubilized in 1% SDS in the presence of protease inhibitors. The suspension was spun at 100,000 g to remove nonsolubilized material, and the top two-thirds of the supernatant were used for further experiments. Protein concentration was measured as in the preceding text.

Immunoblot analysis

Membrane fractions (P3) were solubilized in 1% SDS, 20 mM Tris pH 6.8. After mixing with loading buffer (8% SDS, 40% glycerol, 240 mM Tris-HCl, pH 6.8) and heating for 10 min at 85°C, samples were electrophoresed in a 10% SDS polyacrylamide gel (Harlow and Lane 1988). The electrophoresed proteins were transferred onto a nitrocellulose filter (Bio-Rad), and incubated with Kv3.1a- or Kv3.1b-purified antibodies (1:200 and 1:2,000 dilution, respectively), followed by horseradish peroxidase-linked anti-rabbit secondary antibodies prepared in donkey (Promega, Madison, WI). Bound antibodies were detected using chemiluminescence with an ECL detection kit (Pierce, Rockford, IL).

Sequential co-immunoprecipitations

Before immunoprecipitation, 250 µl solubilized cerebellar membranes (~250 µg protein) in 1% Triton X-100 were precleared for 1 h at 4°C with protein A-Sepharose 4B beads (Zymed Laboratories, South San Francisco, CA). After removing the beads, the extracts were incubated for 2-3 h at 4°C with either Kv3.1a or Kv3.1b antibodies previously bound to protein A-Sepharose beads. At the end of the incubation period, the beads were pelleted (1,000 g for 30 s) and the supernatant was transferred to fresh protein A-Sepharose beads pre-bound to Kv3.1a or Kv3.1b antibodies for two further rounds of immunoprecipitation. After the third round of immunoprecipitations, the supernatant from the three Kv3.1a immunoprecipitations was incubated with Kv3.1b pre-bound beads, while the supernatant extracted for three times with Kv3.1b was incubated with Kv3.1a pre-bound beads; both overnight at 4°C. The beads were pelleted as in the preceding text.

In all cases, complexed beads were collected and washed four times by centrifugation/resuspension with extraction buffer (50 mM Tris-HCl, pH 7.4, 0.15 M NaCl, 1 mM EDTA, 1% Triton X-100]. Bound proteins were then extracted by adding sample buffer (50 mM Tris-HCl, pH 6.8, 5% 2-mercaptoethanol, 20% glycerol, 1.5% SDS), heated for 10 min at 90°C, and processed for immunoblotting as described in the preceding text.

Immunohystochemistry

Adult male C57BL6 mice were deeply anesthetized with pentobarbital sodium (Nembutal; 100 mg/kg ip) and perfused transcardially with a saline solution containing 1U/ml heparin sulfate followed by 4% formaldehyde generated from paraformaldehyde in 0.1 M sodium phosphate buffer, pH 7.4. Brains were postfixed in the same fixative for 1 h and cryoprotected in 30% sucrose at 4°C overnight. Free-floating sections (30 µm), obtained in a freezing microtome, were washed for 30 min in 0.1 M sodium phosphate-buffer saline (PBS), and preincubated with 1% bovine serum albumin (BSA, fraction V) and 0.2% Triton X-100 in 0.1M PBS. The sections were then washed in PBS containing 0.5% BSA (PBS-BSA) and incubated overnight with Kv3.1a (1:100) or Kv3.1b (1:1,000) antisera. After washing in PBS-BSA, the sections were incubated for 1 h with biotinylated anti-rabbit IgG (1:200, Vector Laboratories, Burlingame, CA). The tissue was then washed and incubated for 1 h with the avidin-biotin horseradish peroxidase complex according to Vectastain Elite ABC kit instructions (Vector Laboratories). The antigens were visualized by reaction with the chromogen 3,3'-diaminobenzidine-tetrahydrochloride (DAB, Aldrich) and hydrogen peroxidase for 6 min. Sections were mounted on gelatin-coated slides, dehydrated through graded ethanols, and coverslipped with Permount (Fisher Scientific, St. Louis, MO). The mouse brain atlas by Franklin and Paxinos (1997) and the book by Paxinos (1985) were used as guides to identify CNS neuronal populations and axonal projections.

For immunofluorescence the sections (35-50 µm) were washed with PBS and incubated for 1 h in a blocking buffer (10% normal goat serum, 1% BSA, 0.2% cold fish gelatin, 0.2% Triton X-100 in PBS) to minimize non-specific binding. The tissue was then incubated with primary antibodies in 0.1× strength blocking buffer (working buffer) for 12-20 h at 4°C. For double-labeled sections a primary rabbit (anti-Kv3.1a or anti-Kv3.1b) and a primary mouse antibody [anti-parvalbumin (PV) (Sigma) or anti-SMI-31 (Stemberger Monoclonals, Lutherville, MA)] were added simultaneously. After several washes in working buffer, secondary antibodies were applied for 60 min at room temperature. The following antibody concentrations were used: antibodies against: Kv3.1a at 1:50, Kv3.1b at 1:200, parvalbumin at 1:400, SMI-31 at 1:1,000. Primary antibodies were detected using Cy3-conjugated goat anti-rabbit IgG and Cy2-conjugated goat anti-mouse IgG (Jackson ImmunoResearch, West Grove, PA). Following additional washes, the sections were mounted onto glass slides and coverslipped with a polyvinyl alcohol-glycerol mountant with 2% 1,4-diazabicyclo-[2,2,2]octane (Goslin and Banker 1991). To visualize the fluorescence, we used a Zeiss Axiovert 35M inverted microscope with a laser scanning confocal attachment (MRC-1024; Bio-RAD Laboratories, Cambridge, MA) and a krypton/argon mixed gas laser. Images were collected digitally using either a ×40 oil (n.a. = 1.3) or ×63 (n.a. = 1.4) oil objective and transferred to a graphics program (Adobe Photoshop 5.0).

Electron microscopy

After immunostaining with DAB as described in the preceding text, stained sections were fixed in 2% glutaraldehyde in PBS for 10 min, osmicated in 1% osmium tetroxide in PBS for 30 min, dehydrated in an ascending series of acetone, and embedded in Durcupan ACM resin (Electron Microscopy Sciences, Fort Washington, PA) between glass slides treated with a liquid release agent (Mould Release Compound, Electron Microscopy Sciences). The embedded slices were viewed directly through the plastic with the light microscope and thin (80 nm) or thick (0.5-1 µm) sections were cut on a Reichert Ultracut using a glass or diamond knife. Photographs were either collected on a JEOL 100CX electron microscope at 80-100 keV (thin sections) or on a JEOL JEM-4000EX intermediate voltage microscope at 400 keV (thick sections).

Electron tomography

Three-dimensional volumes of labeled granule cells in the cerebellar granule cell layer were obtained by tomographic reconstruction derived from single-axis tilt series (Frank et al. 1987; Perkins et al. 1997). Thick sections (0.5-1 µm) were cut with a glass knife and collected onto 100 mesh clamshell grids. After washing in ddH₂O, 10-nm colloidal gold was applied to the section surface to serve as fiduciary cues for subsequent alignment of images. Images were typically obtained over a range of +/60° using film at a magnification of ×8,000. Image processing was performed with the SUPRIM software suite (Schroeter and Bretaudiere 1996) using Silicon Graphics workstations. Each negative of the tilt series was digitized with a 14-bit cooled CCD camera (Photometrics) at a resolution of 10 nm/pixel. The 1,024 × 1,024 pixel images were aligned using the colloidal gold particles as fiducial marks, which were selected manually with the program FIDO (Soto et al. 1994). The final volume was rendered and viewed using ANALYZE AVW, a program developed for the measurement of three-dimensional volumes (Robb and Barcillot 1990). Segmentation was performed by manually tracing the structures of interest using the program Xvoxtrace, developed at NCMIR by Stephan Lamont (Perkins et al. 1997).

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Generation and characterization of Kv3.1a-specific antibodies

The two known alternative-spliced products of the Kv3.1 gene differ only in their C-terminal sequence. The last 84 residues in Kv3.1b are replaced by a short 10 amino-acid sequence in Kv3.1a (Luneau et al. 1991a; Rudy et al. 1999; Vega-Saenz de Miera et al. 1994). Kv3.1b-specific antibodies were raised in rabbits immunized with a KLH-linked peptide containing the last 19 residues of the Kv3.1b-specific C-terminus (Weiser et al. 1995). Attempts to raise antisera using a KLH- or BSA-linked peptide containing only the last 10 residues of Kv3.1a as an antigen failed to produce antibodies of sufficient titer. Therefore to prepare good quality antibodies that specifically recognize the Kv3.1a isoform, we immunized rabbits with a primary peptide containing residues C488 to G508 of Kv3.1a (Fig. 1A, bold sequence). This peptide (Ac-CPLAQEEILEINRAGRKPLRG-NH₂) contains 14 residues (CPLAQEEILEINRA) prior to the splice junction, and hence common to both Kv3.1a and Kv3.1b, followed by the first 7 residues (GRKPLRG, the portion predicted to be most antigenic) of the alternative-spliced C terminus of Kv3.1a (Fig. 1A). The resulting antisera reacted strongly with Kv3.1a proteins but displayed cross-reactivity with Kv3.1b proteins. We reasoned that antibodies in the antisera reacting specifically with Kv3.1a proteins could be directed to epitopes present in the unique Kv3.1a carboxyl end and/or epitopes that may include the boundary between the splice junction and the Kv3.1a C-terminus. To purify these antibodies, we performed affinity chromatography using a second peptide (C[AHX]NRAGRKPLRG) in which aminohexanoic acid (AHX) linkers preceded a peptide containing three residues prior to the splice point and the first seven residues of the Kv3.1a-specific sequence (underlined in Fig. 1A).

A series of tests was performed to evaluate both the quality and the specificity of the antibodies (Fig. 1, B-E). Immunoblots of membrane extracts from HEK-293T cells transfected either with Kv3.1a or Kv3.1b cDNA (Fig. 1B) were incubated with Kv3.1a (Fig. 1B, left) or Kv3.1b antibodies (Fig. 1B, right). The Kv3.1a antibodies recognized two bands of ~60 and 95 kDa in immunoblots from Kv3.1a-transfected cells and did not detect any proteins in immunoblots of extracts from Kv3.1b-transfected cells. On the other hand, the Kv3.1b antibodies detected two bands of ~70 and 105 kDa only in the lane corresponding to extracts from Kv3.1b-transfected cells (Fig. 1B, right). With both antibodies, the labeling was absent if the immunoblots were treated with antibodies preincubated with an excess of the respective antigenic peptide (data not shown). The smaller polypeptides recognized by the Kv3.1a and Kv3.1b antibodies are similar in size to that predicted for the core Kv3.1a (~58 kDa) and Kv3.1b (66 kDa) polypeptides. The largest bands probably represent glycosylated forms.

To further establish the specificity of the Kv3.1a and Kv3.1b antibodies, membrane fractions obtained from both wild-type (+/+) and knockout (/) Kv3.1 mouse brain (Ho et al. 1997) were assayed for the detection of Kv3.1a and Kv3.1b proteins (Fig. 1C). Kv3.1a antibodies recognized a single protein of ~75 kDa only in membranes from wild-type (+/+) mice. No bands were detected in immunoblots of membranes from knockout (/) animals (Fig. 1C, left). Similarly, the Kv3.1b antibodies only recognized a single band of ~90 kDa in the wild type but not in the knockout (Fig. 1C, right). The size of this band is in agreement with the molecular weight previously reported in rat brain (Weiser et al. 1995). In both cases, the labeling was prevented by pre-adsorption of the antibodies with the appropriate antigenic peptide (data not shown). It is likely that these proteins correspond to the glycosylated forms of the Kv3.1a and Kv3.1b subunits because their molecular weights are somewhat larger than those predicted from the amino acid sequence of the core polypeptides.

The antibodies were also highly specific and performed well in immunohistochemistry. Immunofluorescence experiments of transfected cells showed that each Kv3.1 antibody immunostained only HEK-293T cells transfected with the appropriate cDNA: The Kv3.1a antibodies labeled Kv3.1a-transfected cells but showed no cross-reactivity toward Kv3.1b-transfected cells (Fig. 1D, top); similarly, the Kv3.1b antibodies were specific for Kv3.1b-transfected cells (Fig. 1D, bottom). In each case, labeling was prevented by pre-adsorption of the antibodies with the respective antigenic peptide (data not shown).

Moreover, all the immunostaining produced by the Kv3.1a or Kv3.1b antibodies in mouse brain tissue was abolished in Kv3.1 knockout mice (Fig. 1E). Together these data show that the antibodies are specific for each of the two Kv3.1 gene products and do not cross react with other brain components.

Distribution of Kv3.1a and Kv3.1b proteins in mouse brain

Northern blot analysis and RNAse protection have shown that Kv3.1b mRNAs are far more abundant than Kv3.1a transcripts in adult brain (Perney et al. 1992; Weiser et al. 1994). It is difficult to obtain quantitative information using immunohistochemistry, particularly when comparing the immunoreactivity produced by two different antibodies. Nevertheless, consistent with the results of RNA analysis, we found much weaker staining of mouse brain with the Kv3.1a than with the Kv3.1b antibodies (Fig. 2).

View larger version (89K): [in this window] [in a new window]   Fig. 2. Immunohistochemical localization of Kv3.1a- and Kv3.1b-spliced isoforms in mouse brain. Consecutive sagittal and coronal sections from a wild-type mouse were immunostained with Kv3.1a (A and B) or Kv3.1b (C and D) antibodies. The Kv3.1a subunit was distributed throughout the brain and was especially prominent in the olfactory bulb, hippocampus, and cerebellum. The Kv3.1b isoform was expressed prominently in the cerebellar cortex and brain stem, as well as in the reticular thalamus, hippocampus, and neocortex. Ia, layer Ia of the olfactory cortex; AON, anterior olfactory nucleus; CA1, CA2, CA3, CA1-CA3 fields of Hammons horn; CPu, caudate putamen; Cu, cuneate nucleus; Cx, cortex; DG, dentate gyrus; Gl, glomerular layer of the olfactory bulb; GP, globus pallidus; Gr, granule cell layer of the cerebellum; IC, inferior colliculus; LH, lateral hypothalamic area; LHb, lateral habenular nucleus; LOT, lateral olfactory tract; MD, mediodorsal thalamic nucleus; Mi, mitral cell layer of the olfactory bulb; Mol, molecular cell layer of the cerebellum; PC, Purkinje cell layer of the cerebellum; Pn, pontine nucleus; Po, posterior thalamic nuclear group; Rt, reticular thalamic nucleus; SNR, substantia nigra, reticular part; STh, subthalamic nucleus; SuG, superficial gray, superior colliculus; VP, ventral pallidum; VPL, ventral posterolateral thalamic nucleus; ZI, zona incerta. Scale bar, 2.2 mm.

Analysis of immunohistochemically stained consecutive sections from wild-type mouse brain showed that, overall, the two Kv3 isoforms were expressed prominently in similar brain regions. As previously reported in rat (Sekirnjak et al. 1997; Weiser et al. 1995) Kv3.1b immunoreactivity was most prominent in the cerebellar cortex (Fig. 2C). Other regions with strong labeling for Kv3.1b were the hippocampus, the reticular thalamic nucleus (Rt), the neocortex and several structures in the brain stem (Fig. 2, C and D). Consistent with observations on the distribution of splice version-specific mRNA transcripts (Perney et al. 1992), Kv3.1a immunostaining was seen in similar areas, although signals were weaker for this isoform, particularly in the brain stem (Table 1; also compare Fig. 2, A and C). On the other hand, there were a few structures, most notably the olfactory bulb, where Kv3.1a immunoreactivity was stronger. In fact, while the olfactory bulb was one of the areas showing highest levels of Kv3.1a protein, this structure expressed Kv3.1b weakly (compare Fig. 2, A and C; Table 1).

Differential subcellular distribution of Kv3.1a and Kv3.1b in the mouse brain

While neuronal somata in many brain areas were strongly labeled with the Kv3.1b antibodies, as observed in previous studies in rat (Du et al. 1996; Perney and Kaczmarek 1997; Sekirnjak et al. 1997; Weiser et al. 1995), in most brain areas, the staining produced by Kv3.1a antibodies was diffuse and restricted to the neuropil with little to no somatic staining. This suggested that Kv3.1a was localized mainly to neuronal processes. For example, in the Rt, a structure consisting mainly of a single population of large GABAergic neurons, and one of the areas expressing Kv3.1 mRNAs most prominently (Perney et al. 1992; Weiser et al. 1994), Kv3.1b antibodies labeled the somata (black arrowheads) and the proximal dendrites (white arrowheads) of Rt neurons (Fig. 3B). The labeled cells are immersed in a strongly labeled neuropil in which some individual fibers can be discerned (Fig. 3B, black arrows). Within the diffuse fiber-like staining, there were also numerous Kv3.1b-stained puncta, which presumably correspond to labeled terminals. The stained fibers form bands, which run in the mediolateral direction, that is, perpendicular to the long axis of the Rt, and therefore perpendicular to the orientation of Rt neurons dendritic arbors (Ohara and Havton 1996; Scheibel and Scheibel 1966), suggesting expression in axonal processes. Unlike Kv3.1b there was no obvious Kv3.1a staining of cell somata or primary dendrites (Fig. 3A). Kv3.1a antibodies only stained the fiber bands and puncta (Fig. 3A). This pattern is consistent with the expression of Kv3.1a proteins being restricted to axonal processes in the Rt.

View larger version (150K): [in this window] [in a new window]   Fig. 3. Differential subcellular localization of Kv3.1a and Kv3.1b in the reticular thalamus and the substantia nigra pars reticulata. Expression of Kv3.1a (A and C) and Kv3.1b (B and D) in consecutive sections of the reticular thalamic nucleus (A and B) and the substantia nigra reticulata (C and D). Kv3.1a immunoreactivity was restricted to axons and terminals in the reticular thalamus while Kv3.1b immunoreactivity was detected both in axons and in neuronal cell bodies (dark arrowheads in B) and proximal dendrites (white arrowheads in B). Analogously, Kv3.1a immunoreactivity (C) in the substantia nigra was diffuse and limited to the neuropil (see higher magnification in inset in C) while Kv3.1b immunoreactivity (D) was detected in both cell bodies (arrowhead in D points to the same neuron in the main panel and in the higher magnification inset) and processes. d, dorsal; l, lateral; m, medial; SNC: substantia nigra, pars compacta; v, ventral. Scale bar: A and B, 24 µm; C and D, 215 µm; insets in C and D, 90 µm.

The type of pattern observed in the reticular thalamus, that is diffuse fiber staining for Kv3.1a in brain regions where Kv3.1b stains, in addition, cell bodies and proximal dendrites, was observed in the substantia nigra pars reticulata (Fig. 3, C and D) and throughout the brain (Table 1). The differences in immunostaining pattern with Kv3.1a and Kv3.1b antibodies were analyzed with more detail in two brain areas, the hippocampus and the cerebellar cortex.

Differential subcellular localization of Kv3.1-spliced isoforms in the hippocampus

Given the laminated organization of different neuronal elements and the well-established cyto-architecture, the clearest area to investigate the differences in subcellular distribution of Kv3.1a and Kv3.1b proteins was the hippocampus. Moreover, in this brain area Kv3.1 mRNA transcripts are only present in subsets of interneurons, including all "basket cells" (Perney et al. 1992; Weiser et al. 1994, 1995), the axons of which produce a clear and recognizable histological pattern (Katsumaru et al. 1988; Kosaka et al. 1987; Sloviter 1989).

The expression pattern of both Kv3.1 isoforms was clearly different in the hippocampal formation (Fig. 4). As previously shown in rat (Du et al. 1996; Sekirnjak et al. 1997; Weiser et al. 1995), Kv3.1b was expressed in basket cell somas throughout the pyramidal cell layers of the hippocampus and along the borders of the stratum (s.) granulosum in the dentate gyrus (arrows). There were also a few stained interneurons in s. oriens and radiatum close to s. pyramidale (arrows, Fig. 4, B and D). In addition to the somatic staining of these neurons, there is staining of a fine net-like structure throughout the s. pyramidale of the CA1-CA3 fields of Ammon's horn (Fig. 4, B and D) and the s. granulosum (Fig. 4F). This staining resembles the characteristic pattern produced by the labeling of the axonal plexus of the basket cells with axonal terminals surrounding pyramidal and granule cells (Celio and Heizmann 1981; Katsumaru et al. 1988; Kosaka et al. 1987; Sloviter 1989).

View larger version (184K): [in this window] [in a new window]   Fig. 4. Differential subcellular immunolocalization of the Kv3.1a and Kv3.1b isoforms in the mouse hippocampus. Consecutive coronal sections of mouse brain were immunostained with Kv3 1a (A, C, and E) or Kv3.1b (B, D, and F) antibodies. Kv3.1a was expressed in processes inside the pyramidal cell layer of the hippocampus (A and C) and in the granule cell layer of the dentate gyrus (E). (Dark arrowheads in E and F mark the margins of the granule cell layer of the dentate gyrus). In the dentate area, there is also prominent Kv3.1a staining of the inner third of the molecular layer (delineated by white arrows) and the hylus (PoDG; E). Kv3.1b was also expressed in the processes inside the pyramidal layers in the CA region (B and D) and the granule cell layer in the dentate gyrus (F); however, Kv3.1b antibodies also prominently stained the cell bodies and proximal dendrites of basket interneurons both in the CA region and in the dentate gyrus (small dark arrows in B, D, and F). The inner third of the molecular layer (delineated by white arrows in F), which was labeled with Kv3.1a antibodies, was not stained with Kv3.1b antibodies. CA1, CA2, CA3, CA1-CA3 fields of Hammons horn; GrDG, granular cell layer, dentate gyrus; hif, hippocampal fissure; LMol: stratum lacunosum moleculare; Or, s. oriens; PoDG, polymorphic layer, dentate gyrus; Py: pyramidal cell layer; Rad: s. radiatum; Scale bar: A and B, 320 µm; C and D, 68 µm; E and F, 125 µm.

In contrast to Kv3.1b, there was little if any staining of cell bodies in the hippocampus with Kv3.1a antibodies. However, there was clear and strong immunostaining of the axonal plexus of the basket cells in the pyramidal cell layers (Fig. 4, A and C) and in the granule cell layer of the dentate gyrus (Fig. 4, A and E). In the dentate gyrus there was also strong, but diffuse staining of the inner third of the molecular layer and the hilus (Fig. 4E).

To further confirm the differential staining of basket cell somata with Kv3.1a and Kv3.1b antibodies and to obtain additional evidence that the fiber staining inside s. pyramidale and granulosum corresponds to the axons of the basket cells, we double-stained mouse sections with Kv3.1a or Kv3.1b antibodies and antibodies to the Ca²⁺-binding protein parvalbumin (PV) and processed the sections for immunofluorescence with secondary antibodies conjugated to two different fluorophores (Fig. 5). PV is a marker of hippocampal basket cells and is present in the cell bodies and axons of these cells (Celio and Heizmann 1981; Freund and Buzsaki 1996; Katsumaru et al. 1988; Kosaka et al. 1987; Sloviter 1989). Basket cell somas and the axonal plexus in s. pyramidale (Fig. 5) and granulosum (data not shown) were stained both with Kv3.1b (Fig. 5B) and PV (Fig. 5D) antibodies and appeared yellow when the fluorescence of the two fluorophores was superimposed (Fig. 5F). In contrast, labeling for Kv3.1a proteins was restricted to the axonal plexus of the basket cells and was double-labeled with PV antibodies (Fig. 5, A, C, and E). Although there are several PV-positive neuronal somata in the field (Fig. 5C), these cells showed little if any somatodendritic Kv3.1a labeling (Fig. 5A). The expression of Kv3.1a (data not shown) and Kv3.1b (Sekirnjak et al. 1997) in axons and terminals in s. pyramidale was also confirmed using immunoelectron microscopy.

View larger version (84K): [in this window] [in a new window]   Fig. 5. Immunolocalization of Kv3.1a and Kv3.1b proteins in parvalbumin (PV)-positive somata and axons in the hippocampus. A, C, and E: dual labeling with Kv3.1a and PV antibodies. Kv3.1a (A) was detected only in the axonal plexus surrounding pyramidal cells. In the same field, antibodies to PV stain the axonal plexus as well as the cell body and proximal dendrites of several interneurons (C). The PV-positive somata are not obviously labeled with Kv3.1a antibodies. Co-localization of Kv3.1a and PV in axons but not in the basket cell somata is apparent when the fluorescence of the two markers is superimposed (E). B, D, and F: Dual labeling with Kv3.1b and PV antibodies. Both Kv3.1b (B) and PV (D) antibodies label basket cell somata and the axonal plexus in s. pyramidale. Co-localization of Kv3.1b and PV in the axonal plexus and basket cell somata is apparent when the fluorescence of the 2 markers is superimposed (F). Scale bar: 55 µm.

Differential subcellular localization of Kv3.1 spliced isoforms in the cerebellar cortex

The cerebellar cortex of the mouse showed the highest levels of both Kv3.1a and Kv3.1b protein expression in the brain (Fig. 2, A and C). Additional evidence of a different subcellular expression pattern of Kv3.1a and Kv3.1b proteins was obtained in this structure where its highly organized cytoarchitecture also facilitated the analysis.

The molecular layer of the cerebellar cortex was strongly labeled by the Kv3.1a (Fig. 6, A and C) and the Kv3.1b antibodies (Fig. 6, B and D). The granule cell layer was also immunolabeled with both antibodies, but the intensity of the label in the molecular layer was stronger than the staining in the granule cell layer, where the mRNA for these proteins is located (Perney et al. 1992; Weiser et al. 1994). The labeling in the molecular layer could not result from protein present in the scattered basket and stellate cells of this layer. Observation of the molecular cell layer at higher magnification (Fig. 6, C and D) suggests that the labeling in this area is due to protein present in the parallel fiber system, the axons of the granule cells. The ascending fibers of the parallel fiber system are labeled and can be clearly seen as they cross the Purkinje cell layer (arrowheads, Fig. 6, C and D).

View larger version (128K): [in this window] [in a new window]   Fig. 6. Immunohistochemical localization of Kv3.1a and Kv3.1b isoforms in the mouse cerebellum. Consecutive coronal sections were stained with Kv3.1a (A, C, and E) or Kv3.1b (B, D, and F) antibodies. Kv3.1a (A and C) and Kv3.1b (B and D) were strongly expressed in the cerebellar cortex. Antibodies to both isoforms produced prominent labeling of the molecular (Mol) and granule cell (Gr) layers. Purkinje cells (PC) were not labeled, but stained parallel fibers (the axons from the granule cells) were stained and are clearly observed as they cross the unlabeled Purkinje cell layer (arrowheads in C and D). Deep cerebellar nuclei [e.g., interposed, lateral, and medial cerebellar nucleus (IntDL, Lat and Med)] were also labeled with both Kv3.1a and Kv3.1b antibodies (A and B, respectively) although with different patterns. Kv3.1a labeling is faint and diffuse and consists of stained neuropil, while Kv3.1b labeling also includes cell body staining (data not shown). Several stained structures in the brainstem associated with these sections are visible in A and B, although the staining with Kv3.1b is much stronger than that with Kv3.1a. E and F: high magnification of the granule cell layer demonstrates that the patterns of staining in this layer differed for Kv3.1a (E) and Kv3.1b (F). Kv3.1a-labeling is punctate and appears to be intracellular or associated with structures present in between granule cells (E; arrows point to some of the punctate labeling that appears to be intracellular) while Kv3.1b-labeling (F) delineates the granule cell's somatic membrane (arrows in F). Asterisks in E and F mark the center of a few granule cell somata. DC, dorsal cochlear nucleus; g, glomerulus; icp, inferior cerebellar peduncle; MVvePC, medial vestibular nucleus; sp5, spinal trigeminal tract; SpVe, spinal vestibular nucleus; VCP, ventral cochlear nucleus. Scale bar: A and B, 600 µm; C and D, 30 µm; E and F, 13 µm.

Both antibodies also stained the granule cell layer, where the cell bodies of the granule cells are located (Fig. 6, A-D). However, the histological appearance of the staining with the two antibodies was obviously different when observed at high magnification. While Kv3.1b staining clearly delineated the somatic membrane of the granule cells (Fig. 6F), Kv3.1a antibodies produced a fine punctate staining apparently inside granule cell somata (Fig. 6E). In addition we observed punctate label that appeared to be located outside the cell bodies of granule cells (Fig. 6E).

We used immunoelectron microscopy to further analyze the subcellular localization of Kv3.1a and Kv3.1b proteins in the cerebellar cortex (Fig. 7). In granule cell somata, ultrastructural analysis showed that the Kv3.1a immunoreaction product was present in intracellular patches (Fig. 7, A, C, D, and E). These patches were often in vesicular organelles close to the Golgi apparatus (g, Fig. 7D) but were also sometimes seen near the plasma membrane (Fig. 7A). These patches probably account for the intracellular punctate staining observed with transmission microscopy. Kv3.1a label was rarely found in somatic plasma membrane or dendrites with the electron microscope. This pattern was quite different than that produced by Kv3.1b antibodies, which produced strong immunoproduct that delineated the plasma membrane of granule cell somata and proximal dendrites (Fig. 7B) (see also Sekirnjak et al. 1997; Weiser et al. 1995). In addition to the intracellular labeling produced by Kv3.1a antibodies, there was clear staining of identified axonal processes emanating from and passing between granule cells (Fig. 7, D and E). In contrast to the somatic labeling, stained axons often showed membrane associated immunoreaction product (arrowheads, Fig. 7D; see also F). Prominently labeled axons were also seen in the molecular layer with Kv3.1a (Fig. 7F) and Kv3.1b (Sekirnjak et al. 1997) antibodies.

View larger version (161K): [in this window] [in a new window]   Fig. 7. Electron micrographs of Kv3.1a and Kv3.1b immunoreactivity in the cerebellar cortex. A: Kv3.1a-labeled granule cells showing patches of immunoreaction product in the cytoplasm, some of the patches are close to the plasma membrane. B: Kv3.1b-labeled granule cell showing immunoreactivity along the plasma membrane. C: Kv3.1a-labeled granule cell showing intracellular labeling (arrow). D: Kv3.1a-labeled granule cell showing immunoreactivity in a vesicular organelle associated with the Golgi apparatus (g). The axon emanating from the cell is also labeled and shows immunoreaction product along its plasma membrane (arrowheads). E: image of Kv3.1a immunoreactivity in axons coursing through the granule cell layer (arrows). F: Kv3.1a immunoreactivity in the molecular layer of the cerebellar cortex. Arrows point to axons where labeling of the plasma membrane is apparent. n, nucleus. Scale bar: 0.5 µm.

Some of the patches of Kv3.1a immunoreactivity were seen close to the plasma membrane (Fig. 7A). Granule cells are small neurons and have a thin layer of cytoplasm. This can complicate the interpretation of labeling patterns using immunoperoxidase. Therefore to further confirm the apparent differences of Kv3.1a and Kv3.1b immunolabeling close to the plasma membrane of granule cell somata, we performed three-dimensional reconstruction of serial sections analyzed with electron microscopy. The tomograms shown in Fig. 8 clearly illustrate the differences in protein localization in granule cell perikarya. Most Kv3.1b staining is associated with the plasma membrane. It forms ring structures as previously suggested (Sekirnjak et al. 1997). In contrast, the tomograms show that the Kv3.1a protein in granule cell somata that is close to the plasma membrane is indeed in intracellular patches (Fig. 8). Cerebellar granule cell somas seem to contain more intracellular Kv3.1a protein than other neurons also expressing the protein in the axonal compartment. This might be related to the specific biosynthetic requirements necessary to feed channel proteins into the extensive parallel fiber system by cells having small amounts of cytoplasm.

View larger version (107K): [in this window] [in a new window]   Fig. 8. Three-dimensional tomographic reconstructions of the labeling for Kv3.1a and Kv3.1b in individual granule cells in the cerebellar cortex. Top: computer generated slices through granule cells labeled with Kv3.1a (left) or Kv3.1b (right) antibodies. Bottom: 3-dimensional surface reconstructions of the plasma membrane and labeled regions of the same cells pictured in the top panels. Kv3.1a-labeling is pictured on the left and Kv3.1b-labeling is pictured on the right. The membrane is shown in green (Kv3.1a) and blue (Kv3.1b); the channel protein immunolabel is shown in yellow (Kv3.1a) and pink (Kv3.1b). In contrast to the patchy labeling seen with Kv3.1a, which is mostly intracellular, labeling for Kv3.1b is distributed in regularly spaced bands around the cell membrane. A portion of the Kv3.1a-labeled cell was cut off to illustrate that most of the immunoreaction product is intracellular. Scale bar: 1 µm.

Exceptions to the predominantly axonal localization of Kv3.1a proteins: Kv3.1a proteins in somas and dendritic processes

So far, in all the brain areas analyzed, Kv3.1a was found predominantly in fine-caliber processes that seem to correspond to axonal processes and in puncta suggestive of presynaptic terminals. Using electron microscopy, we confirmed this interpretation in the hippocampus and cerebellum. In cells showing clear somatic Kv3.1a staining, as in cerebellar granule cells, the protein appears to be localized in intracellular organelles and not in somatic membrane. We found two exceptions to this staining pattern: in the olfactory bulb (Figs. 2, 9, and 10) and in neurons in the mesencephalic trigeminal nucleus (Fig. 11).

View larger version (118K): [in this window] [in a new window]   Fig. 9. Localization of Kv3.1a, but not Kv3.1b, in somas, dendrites and axons of mitral cells of the olfactory bulb. A: Kv3.1a antibodies prominently stained the cell bodies of the mitral cells (arrows) in the mitral cell layer (Mi). Stained primary dendrites are evident at higher magnification (arrowheads in inset). There is also prominent staining of the external plexiform layer (Epl), which is particularly strong in the two-thirds closest to the Mi, suggesting high levels of Kv3.1a protein in the secondary (or lateral) dendrites of the mitral cells. Diffuse staining of the glomerular layer (Gl) is also apparent. Kv3.1a staining of the proximal portion of the mitral axons in the internal plexiform layer (IPl) was also prominent. B: Kv3.1b was present in some tufted and periglomerular cell bodies. There is also diffuse staining of the IPl, which is much weaker than that seen with Kv3.1a antibodies. GrO, granular cell layer. Scale bar: A and B, 85 µm; inset, 50 µm.

View larger version (80K): [in this window] [in a new window]   Fig. 10. Immunolocalization of Kv3.1a in mitral cells with the electron microscope. A: in mitral cell somata, Kv3.1a labeling was mostly intracellular and often associated with the rough endoplasmic reticulum (arrow) close to the nucleus (n). B and C: membrane Kv3.1a staining (arrows) is evident in large and fine dendritic processes (d) in the external plexiform layer of the olfactory bulb. Scale bar: 1 µm.

View larger version (86K): [in this window] [in a new window]   Fig. 11. Kv3.1a but not Kv3.1b proteins were expressed in the somata of a subpopulation of Me5 trigeminal neurons in the brain stem. Immunodetection of Kv3.1a (A), phosphorylated neurofilaments with SMI-31 monoclonal antibody (B) and Kv3.1b (C) in sagittal sections of the mouse brain. Note immunostained Me5 neurons (arrows) with Kv3.1a and SMI-31 but not with Kv3.1b antibodies. D-F: confocal microscope images of a dual labeling immunofluorescence showing the localization of Kv3.1a in the somatic membrane surrounding the cell body (D) in trigeminal cells dual stained with SMI-31 antibody (E). The images with both antibodies are shown superimposed in (F); Kv3.1a immunofluorescence is shown in red and SMI-31 in green. Cer, cerebellum; Me5, mesencephalic trigeminal nucleus. Scale bar: A-C, 200 µm; inset in A, 43 µm; D-F, 30 µm.

In the olfactory bulb Kv3.1a is expressed in mitral cell somas, dendrites, and axons

In situ hybridization showed prominent Kv3.1 gene expression in mitral cells of the olfactory bulb (Weiser et al. 1994). It was therefore surprising that there was no Kv3.1b protein in these cells in rat brain (Weiser et al. 1995). In this study, we also did not detect Kv3.1b protein in mitral cells in mouse (Fig. 9B). However, the Kv3.1a isoform was expressed at high levels in these cells. All the domains of the mitral cells were strongly stained, including cell bodies and proximal and secondary dendrites (Fig. 9A), producing strong staining of the external plexiform layer (EP1), particularly heavy in the two-thirds proximal to mitral cell somas (Fig. 9A), where the lateral dendrites of the mitral cells are located. The prominent Kv3.1a staining of dendritic processes in the EP1 was confirmed with immunoelectron microscopy (Fig. 10).

Finally, the axons of the mitral cells also expressed abundant Kv3.1a proteins (Figs. 2, A and B, and 9A). The staining was particularly prominent in the initial segments of these axons as they emerge from mitral cells and concentrate in the internal plexiform layer (IP1, Fig. 9A) and in their terminal fields in layer Ia of the anterior olfactory nucleus (AON), piriform cortex, and olfactory tubercle (Fig. 2, A and B). This is the layer where mitral cell axons, upon leaving the lateral olfactory tract (LOT) and losing their myelination, contact the dendrites of olfactory cortex pyramidal neurons (Gracey and Scholfield 1990; Ojima et al. 1984; Shipley et al. 1995; Switzer et al. 1985). The staining of the terminal fields of mitral cell axons was much stronger than the staining of the axons within the LOT itself (Fig. 2A), indicating that Kv3.1a channels are more abundant in unmyelinated axons as observed for other Kv3 proteins (Moreno et al. 1995; Sekirnjak et al. 1997; Weiser et al. 1995). Kv3.1b antibodies stained mitral cell axons much more weakly (Figs. 2C and 9B). Moreover, layer Ib of the olfactory cortex, which contains synapses between recurrent collateral axons and dendrites of pyramidal neurons, was not stained with either Kv3.1a or Kv3.1b consistent with little expression of Kv3.1 transcripts in cortical pyramidal neurons.

Somatic staining of mesencephalic trigeminal neurons

Kv3.1a channel protein was also prominently expressed in the somas of a subpopulation of neurons of the mesencephalic trigeminal nuclei in the brain stem (Fig. 11, A and inset). In these cells, Kv3.1a staining delineated the plasma membrane (Fig. 11D), and this membrane staining was confirmed by electron microscopy immunohistochemistry (data not shown). However, Kv3.1b was not detected in these cells (Fig. 11C).

Interestingly, these cells represent a rare population of CNS neurons that are thought to undergo aberrant phosphorylation of neurofilaments in the perikaryon, resulting in somatic labeling by an antibody against phosphorylated neurofilaments (SMI-31), which otherwise stains only axons in normal brain (Klosen and Van den Bosch de Aguilar 1994; Poltorak and Freed 1987). Utilizing the SMI-31 antibody, we could demonstrate the presence of Kv3.1a immunoreactivity in cells expressing the phosphorylated neurofilament epitope (Fig. 11, D-F), although the SMI-31 antibody staining is seen throughout the cytosol, while the Kv3.1a labeling is mainly membrane-associated.

Most Kv3.1a protein in the cerebellum is in heteromeric complexes with Kv3.1b subunits

Kv3.1a and Kv3.1b proteins overlapped in the axonal compartment of several neuronal populations in the brain. Because any given Kv3 protein can form heteromeric complexes with other Kv3 proteins (Rudy et al. 1999; Vega-Saenz de Miera et al. 1994; Weiser et al. 1994), it is possible that both Kv3.1 isoforms are part of the same heteromultimeric channels in these axons. We explored this hypothesis using co-immunoprecipitation assays from non-denaturing detergent extracts of mouse cerebellar membrane fractions. Kv3.1a antibodies immunoprecipitated both Kv3.1a (data not shown) and Kv3.1b (Fig. 12, lane 4) proteins. Similarly, Kv3.1b antibodies immunoprecipitated both isoforms (Fig. 12, lanes 1 and 6). Moreover, most of the Kv3.1a protein present in the extract was immunoprecipitated by Kv3.1b antibodies such that addition of Kv3.1a antibodies to a membrane extract previously immunoprecipitated with Kv3.1b antibodies did not cause much more Kv3.1a protein to precipitate (Fig. 12, lanes 1-3). In contrast, the reciprocal experiment showed that only a fraction of the Kv3.1b protein present in the extract could be precipitated by Kv3.1a antibodies (Fig. 12, lanes 4-6), consistent with the lack of co-localization of Kv3.1a and Kv3.1b proteins in the plasma membrane of granule cell somata.

View larger version (63K): [in this window] [in a new window]   Fig. 12. In cerebellar membranes most Kv3.1a protein is in complexes containing Kv3.1b subunits. Protein was extracted from membrane preparations of cerebellar tissue with non-denaturing detergents, and immunoprecipitation (IP) reactions of the protein samples performed with the indicated antibodies. The immunoprecipitation products were recovered and separated using SDS-PAGE, transferred to nitrocellulose membranes and probed with the indicated antibodies (BLOT). Lanes 1-3: the protein extract was immunoprecipitated with Kv3.1b antibodies, and the immunoprecipitation products were probed with Kv3.1a antibodies (lane 1). The supernatant of this immunoprecipitation reaction was re-precipitated with the Kv3.1b antibodies, and the IP products were probed again with the Kv3.1a antibodies (lane 2). Note that most of the Kv3.1b-immunoprecipitable Kv3.1a protein was recovered during the first immunoprecipitation. Following a 3rd immunoprecipitation with Kv3.1b antibodies, the supernatant was immunoprecipitated with Kv3.1a antibodies and the IP products probed also with Kv3.1a antibodies (lane 3). Little additional Kv3.1a protein was detected, indicating that most of the available Kv3.1a protein in the extract was already precipitated by the Kv3.1b antibodies. Lanes 4-6: two sequential immunoprecipitation reactions with Kv3.1a antibodies were performed and the IP products probed for Kv3.1b proteins (lanes 4 and 5). Most of the Kv3.1a-immunoprecipitable Kv3.1b protein was precipitated during the 1st IP reaction (compare lanes 4 and 5). Following a 3rd immunoprecipitation with Kv3.1a antibodies, the supernatant was immunoprecipitated with Kv3.1b antibodies and the IP products probed with Kv3.1b antibodies (lane 6). When the supernatant of the 3rd Kv3.1a immunoprecipitation reaction was immunoprecipitated with Kv3.1b antibodies a large amount of Kv3.1b protein was recovered (lane 6) indicating that a large fraction of the Kv3.1b protein in the extract is not associated with Kv3.1a proteins.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Role of the alternative splicing of C-terminal sequences of Kv3 K⁺ channel proteins in membrane targeting

Neurons are asymmetric cells consisting of several structural and functional domains, which, to some degree or another, can be electrically and chemically independent. Therefore, the functional implications of any given ion channel will depend on their precise localization on the neuronal surface. Protein localization to different membrane domains depends on various specialized sorting mechanisms as well as on selective exclusion and retention mechanisms such as stabilization by components of the submembrane cytoskeleton or interactions with matrix proteins (Bradke and Dotti 1998; Foletti et al. 1999; Higgins et al. 1997; Ikonen and Simons 1998; Matter 2000; Mattson 1999; Mostov et al. 2000; Trimmer 1999; Yeaman et al. 1999).

The present study revealed a different subcellular distribution pattern of the two alternative-spliced isoforms of the voltage-gated potassium channel Kv3.1 gene in mouse CNS neurons using splice-version-specific antibodies. The results support the view that an important function of the C-terminal alternative splicing of Kv3 genes is to generate isoforms with different membrane localizations. This conclusion is also supported by the observation that different Kv3 subunits are targeted to distinct membrane domains in model polarized epithelial cells, depending on the sequence of their C terminus (Ponce et al. 1997). The C-terminal domains of Kv3 proteins also contain different putative sites for phosphorylation (Vega-Saenz de Miera et al. 1994), and differential modulation by PKC of two heterologously expressed Kv3.2 spliced versions has been reported (McIntosh et al. 1998). Therefore it is possible that targeting of Kv3 subunits to different neuronal regions is associated with channel modulation by different stimuli.

The results showed that, except for mitral cells in the olfactory bulb and mesencephalic trigeminal neurons (see following text), in native neurons the Kv3.1a protein was found almost exclusively in axons and presynaptic terminals. In neurons where there was significant somatic staining with Kv3.1a antibodies, as in cerebellar granule cells, the protein was usually in intracellular organelles, presumably in traffic toward its axonal localization. This suggests that the Kv3.1a subunit contains axonal targeting determinants. Consistent with this idea, we have observed that Kv3.1a proteins are specifically targeted to the apical domain in MDCK cells (Ozaita and Rudy, unpublished observations); this is thought to be equivalent in many instances to the axonal compartment in neurons (Dotti and Simons 1990; Higgins et al. 1997; Jareb and Banker 1998). Kv3.1a and Kv3.1b differ only in their C-terminal sequence (Luneau et al. 1991a; Rudy et al. 1999; Vega-Saenz de Miera et al. 1994), suggesting that their differential subcellular localization is regulated by their C-terminal sequence. One possibility is that the C-terminal sequence of Kv3.1a contains an axonal targeting signal. Alternatively, it is possible that the C-terminal sequence controls the function of a targeting signal present somewhere else in the protein.

Voltage-gated K⁺ channels of the Kv family are tetrameric and can be formed by association of four identical subunits or by combinations of different Kv subunits of the same subfamily (reviewed in Coetzee et al. 1999). The axonal targeting property of Kv3.1a proteins may confer to these subunits the ability to direct heteromeric Kv3 channels to the axonal compartment. This could explain the presence of Kv3.1b proteins in axons without proposing that this isoform also has axonal targeting capabilities of its own. Other Kv3 proteins having axonal targeting determinants may carry Kv3.1b proteins to axons in neurons that express low levels of Kv3.1a proteins as may be the case in several neuronal populations in the brain stem (see Fig. 2). Isoforms of the Kv3.3 gene are particularly good candidates because there is an extensive overlap between Kv3.1 and Kv3.3 gene expression in the brain stem (Weiser et al. 1994). Accordingly, channels containing Kv3.1b proteins might be somatodendritic when present in complexes not having Kv3.1a proteins (or other Kv3 isoforms with axonal targeting capabilities) and axonal when in hetero-oligomers containing these axonal targeting subunits. Thus the main function of Kv3.1a proteins might be to direct Kv3 channels to axons and terminals. This hypothesis is consistent with the results from the co-immunoprecipitation studies, which showed that most of the Kv3.1a protein in the cerebellum is associated with Kv3.1b subunits, but a significant fraction of the Kv3.1b protein is not in complexes containing Kv3.1a (Fig. 12). It is also supported by experiments in MDCK cells showing that Kv3.1a proteins can redirect to the apical domain Kv3 proteins that on their own are only expressed in the basolateral membrane (Ozaita and Rudy, unpublished observations). On the other hand, it is possible that both Kv3.1a and Kv3.1b contain axonal targeting signals in the common regions of the proteins and that the C-terminal domains regulate somatodendritic targeting. For example, the Kv3.1a C-terminal sequence could contain a somato-dendritic exclusion signal not present in the Kv3.1b C-terminal sequence. The present study demonstrating the differential distribution of the two Kv3.1 isoforms in brain neurons justifies mutagenesis targeting experiments to try to distinguish between these possibilities. Because Kv3.1a and Kv3.1b proteins differ only in their C-terminal sequence, they provide an excellent system to investigate the nature of targeting signals in neurons.

Kv3.1a proteins in mitral cell dendritic processes

The expression of Kv3.1a proteins throughout the mitral cell dendritic arbor is the only presently known case of prominent localization of Kv3 proteins in fine dendrites in native mammalian CNS neurons. Kv3.1b (Du et al. 1996; Sekirnjak et al. 1997; Weiser et al. 1995), Kv3.2 (Chow et al. 1999; Moreno et al. 1995), and Kv3.4 (Veh et al. 1995) proteins, when expressed somatically, can be detected in the initial segments of primary dendrites but not in fine dendritic processes. Although Kv3 proteins have not been demonstrated in mammalian dendrites, Rashid et al. (2001) found that Kv3.3 ortholog proteins are prominent in the dendrites of pyramidal cells of the electrosensory lobe (ELL) of an apteronotid weakly electric fish.

The dendritic localization in mitral cells of what otherwise would appear to be an axonal membrane protein is quite interesting and most likely reflects the fact that mitral cell dendrites are unusual in having both a dendritic and an axonal character. Mitral cell dendrites participate in reciprocal dendro-dendritic synapses, release neurotransmitter (glutamate), and thus contain presynaptic vesicles and other elements of presynaptic terminals (Isaacson and Strowbridge 1998; Nicoll 1969; Rall et al. 1966; Shepherd and Greer 1998). Therefore these dendrites have to be able to capture proteins that are normally targeted to axons and must include the molecular machinery necessary for the trafficking and retention of axonal proteins.

The transmission of action potentials in mitral cell dendrites is also different from that in dendrites from most neurons (Bischofberger and Jonas 1997). In most neurons, the action potential is attenuated and prolonged as it invades the dendrites. However, in mitral cells the large amplitude and fast time course of the action potential are maintained as it back-propagates from the soma into the dendritic tree, a property that is thought to be important for the efficient release of neurotransmitter, and probably depends on the large density of Na⁺ and K⁺ conductances in the dendritic membrane (Bischofberger and Jonas 1997).

Bischofberger and Jonas (1997) observed that a large fraction (>65%) of the K⁺ current in mitral cell dendrites was blocked by 1 mM TEA. Kv3.1a subunits might contribute significantly to this mitral cell dendritic K⁺ current. This would be consistent with the pharmacological observations of Bischofberger and Jonas (1997) because at the dose used by these investigators, TEA blocks only a few K⁺ channel types including Kv3's (Rudy et al. 1999). Kv3 channels have unusual electrophysiological properties and are believed to be specialized for mediating fast action potential repolarization with minimum effects (as compared to other voltage-gated K⁺ channels) on action potential threshold or magnitude (Rudy and McBain 2001). Therefore, Kv3 channels are ideally suited to generate the fast, large-amplitude action potentials propagating in mitral cell dendrites to efficiently trigger glutamate release (Bischofberger and Jonas 1997).

Kv3.1a proteins in mesencephalic trigeminal neurons

The large neurons in the mesencephalic trigeminal nucleus were the only neuronal population in the mouse brain that prominently expressed Kv3.1a proteins in somatic membrane. This is an interesting finding given the observation that these cells are one of the rare examples of aberrant labeling in brain by antibodies against phosphorylated neurofilaments (e.g. SMI-31), which otherwise stain only axons in normal neurons (Klosen and Van den Bosch de Aguilar 1994; Poltorak and Freed 1987). Phosphorylated neurofilaments in the perikarya of normal neurons has been observed only in mesencephalic trigeminal neurons, in some bipolar septofimbrial neurons and in a subset of cells in dorsal root ganglia. Other neurons only express these epitopes somatically following axotomy or in pathological situations such as in Alzheimer's disease (Klosen and Van den Bosch de Aguilar 1994; Poltorak and Freed 1987). Although, to our knowledge it is not known why a few neuronal types have these cytoskeletal singularities, the observations are consistent with the possibility that mesencephalic trigeminal neurons read axonal targeting signals differently than most neurons.

Thus the observation of somatodendritic expression of Kv3.1a in mitral and mesencephalic trigeminal neurons does not necessarily conflict, but in fact might be consistent with the notion that this protein contains axonal targeting signals and contributes to the formation of heteromeric Kv3 channels having an axonal and presynaptic-terminal localization.

Based on pharmacological experiments, Rashid et al. (2001) have suggested that dendritic Kv3.3 channels influence the threshold for burst discharge of neurons in the ELL in electric fish. Interestingly, characteristic bursting behaviors have also been observed in mitral cells (Duchamp-Viret and Duchamp 1993) and in mesencephalic trigeminal neurons (del Negro et al. 1999). Kv3.1a channels may play a role in regulating bursting in these neurons as well (see Rudy and McBain 2001 for a discussion on the role of Kv3 channels in bursting).