Small-conductance Ca2+-activated K+ (SK) channels are important for excitability control and afterhyperpolarizations in vertebrate neurons and have been implicated in regulation of the functional state of the forebrain. We have examined the distribution, functional expression, and subunit composition of SK channels in rat brain. Immunoprecipitation detected solely homotetrameric SK2 and SK3 channels in native tissue and their constitutive association with calmodulin. Immunohistochemistry revealed a restricted distribution of SK1 and SK2 protein with highest densities in subregions of the hippocampus and neocortex. In contrast, SK3 protein was distributed more diffusely in these brain regions and predominantly expressed in phylogenetically older brain regions. Whole-cell recording showed a sharp segregation of apamin-sensitive SK current within the hippocampal formation, in agreement with the SK2 distribution, suggesting that SK2 homotetramers underlie the apamin-sensitive medium afterhyperpolarizations in rat hippocampus.
Molecular cloning of small-conductance Ca2+-activated K+ (SK) channels has revealed three genes, SK1–SK3 (Kohler et al., 1996). When expressed in Xenopusoocytes, the resulting channels are voltage-insensitive and activated by submicromolar intracellular Ca2+(Kohler et al., 1996; Hirschberg et al., 1998). Structure–function analysis revealed that their Ca2+ gating is mediated by constitutive association with calmodulin (Xia et al., 1998; Keen et al., 1999). However, direct interaction of calmodulin with native brain SK channels so far has been demonstrated only for SK3 (Xia et al., 1998).
In many neurons, action potentials are followed by afterhyperpolarizations (AHPs) consisting of several phases, reflecting the activation of different K+ currents. Hippocampal pyramidal cells show three different AHPs: fast, medium (mAHP), and slow (sAHP) (Storm, 1987, 1990; Sah, 1996). Because SK2 and SK3 channels are highly apamin-sensitive (picomolar range), they are believed to contribute to the formation of the apamin-sensitive mAHP component and the corresponding current (Sah, 1996; Vergara et al., 1998; Xia et al., 1998; Stocker et al., 1999). For clarity, we will use the term I aAHP for the apamin-sensitive component of the mAHP current, because other current components [muscarin-sensitive K+ (M)-current, large-conductance Ca2+-activated K+(BK)-channel-mediated current, and hyperpolarization-activated cation current] are also known to contribute to the mAHP (Storm, 1989, 1990;Williamson and Alger, 1990). In contrast to SK2 and SK3 channels, homomeric SK1 channels show a significantly lower apamin sensitivity (nanomolar range). It was therefore suggested that the sAHP is mediated by SK1 channels. However, the findings that the sAHP is resistant to apamin (up to 1 μm; Lancaster and Nicoll, 1987;Storm, 1989; Stocker et al., 1999), whereas SK1 homomeric channels are blocked by this toxin (Shah and Haylett, 2000a; Strobaek et al., 2000;Grunnet et al., 2001a), have, together with other pharmacological and kinetic discrepancies, cast doubt on the proposed causal relationship between SK1 and the sAHP (Shah and Haylett, 2000b; Shah et al., 2001;Faber and Sah, 2002). Meanwhile, a complex alternative splicing pattern of mouse and rat SK1 genes have been discovered (Shmukler et al., 2001), with some splice variants lacking the calmodulin-binding domain.
To determine which SK proteins underlie the various ionic currents and AHPs, detailed information regarding the distributions of the SK subunits is required. However, so far our knowledge of these distributions in the brain has been based on in situhybridization data (Kohler et al., 1996; Stocker et al., 1999; Stocker and Pedarzani, 2000) and a few reports using antibodies directed against SK1 and SK3 protein (Bond et al., 2000; Bowden et al., 2001;Tacconi et al., 2001).
In this study, we have used a combination of immunological, pharmacological, and electrophysiological tools to determine the distribution and functional expression of all three SK channel genes in rat brain. These results may help clarify unsolved discrepancies of functional and pharmacological properties of SK channels.
MATERIALS AND METHODS
[125I]Apamin (2175 Ci/mmol) was obtained from PerkinElmer Life Sciences (Boston, MA), and SuperSignal West dura extended substrate was from Pierce (Rockford, IL). Protein A-Sepharose, sodium cholate, peroxidase-conjugated goat anti-rabbit IgG (used for Western blots), and 3,3′ diaminobenzidine were from Sigma (Munich, Germany). A monoclonal mouse IgG1anti-calmodulin antibody (05-173) was purchased from Upstate Biotechnology (Lake Placid, NY). Horseradish peroxidase-coupled goat anti-rabbit IgG (0448) was purchased from Dako (Glostrup, Denmark). Apamin was bought from Calbiochem (San Diego, CA), and 9-fluorenylmethoxycarbonyl lysine core solid phase support was from NovaBiochem (Laeufelfingen, Switzerland). Cyanogen bromide (CNBr)-activated Sepharose 4B was obtained from Amersham Biosciences (Uppsala, Sweden). The M-channel blocker 10,10-bis(pyridinylmethyl)-9(10H)-anthracenone (XE991) was obtained from DuPont (Billerica, MA). All remaining drugs were from Sigma. Substances for slice electrophysiology were bath-applied by adding them to the superfusing medium.
Antibody production and affinity purification
Polyclonal sera were raised against SK1 protein (GenBank accession number U69885), residue positions 12–29, using the sequence QPLGSGPGFLGWEPVDPE (anti-SK1(12–29)) and 515–532, using the sequence HLTTAAQSPQSHWLPTTA (anti-SK1(515–532)); SK2 protein (GenBank accession number U69882), residue position 538–555, using the sequence RDFIETQMENYDKHVTYN (anti-SK2(538–555)); and SK3 protein (GenBank accession number U69884), residue position 504–522, using the sequence ADTLRQQQQQLLTAFVEAR (anti-SK3(504–522)) (Kohler et al., 1996). Antibodies were raised and affinity-purified as described previously (Knaus et al., 1995).
Immunoblot analysis was performed as described by Knaus et al. (1995) with some minor modifications. Purified hippocampal synaptic plasma membrane vesicles (10–30 μg/lane) or crude oocyte membranes were separated by 10% SDS-PAGE and transferred to polyvinylidene difluoride (PVDF) membranes. The membrane was blocked with 3% BSA, 0.5% (w/v) Triton X-100, and 0.1% (w/v) Tween 20 dissolved in Tris-buffered saline (TBS) for either 120 min at 22°C or 12–15 hr at 4°C. Thereafter, the blots were incubated with anti-SK antibodies (affinity-purified anti-SK1(12–29), 1 ng/μl IgG; affinity-purified anti-SK1(515–532), 2 ng/μl IgG; and crude serum of anti-SK2(538–555) and anti-SK3(504–522), 1:5000 and 1:13000, respectively) or purified mouse monoclonal anti-calmodulin antibody (Upstate Biotechnology; 1:1000) diluted in 3% BSA, 0.5% Triton X-100, and 0.1% Tween 20 for 12 hr at 4°C. Blots were washed three times with 0.5% Triton X-100 and 0.1% Tween 20 in TBS and incubated with affinity-purified horseradish peroxidase-conjugated goat anti-rabbit or anti-mouse IgG for 120 min at 22°C. After washing six times with the buffer described above, blots were developed using the SuperSignal West dura extended substrate detection kit according to the manufacturer's protocol.
Immunocytochemical experiments were performed in slight variations to our previous study (Knaus et al., 1996) using free-floating 40 μm cryosections of 4% (w/v) paraformaldehyde-perfused rat brains. Brain sections were permeabilized for 60 min in TBS, pH 7.4, and 0.4% (w/v) Triton X-100, subsequently blocked in 10% normal goat serum (NGS) in TBS and Triton X-100 for 90 min, and incubated overnight in 3% NGS containing affinity-purified anti-SK1(12–29) antibody at a concentration of 2 ng/μl. Crude antiserum of SK2(538–555) was used at final dilutions of 1:4000–1:9000, whereas anti-SK3(504–522) antiserum was diluted 1:12000–1:15000 or used at a concentration of 1.3 ng/μl for affinity-purified antibody. Three rinses in TBS and Triton X-100 were performed before incubation with secondary antibody (horseradish peroxidase-coupled goat anti-rabbit IgG, 1:400) for 150 min. After three washes with TBS, pH 8 (in mm: 20 Tris-HCl, pH 8, and 150 NaCl), antigen-antibody complexes were visualized by reaction with 3,3′-diaminobenzidine, nickel ammonium sulfate, and H2O2 in TBS, pH 8, at final concentrations of 0.63 mm, 10 mm, and 0.002%, respectively (Wouterlood, 1988). In control sections, nonspecific immunoreactivity was assessed by preadsorbing primary antibodies with a 10 μm concentration of the respective peptide, incubations without the primary antibody, or using preimmune serum (see Fig. 3). After immunostaining, the preparations were dehydrated in an ethanol series, cleared with butylacetate, and mounted in EuKitt (Christine Groepl, Vienna, Austria). The sections were analyzed using aZeiss (Oberkochen, Germany) Axioplan 2 microscope equipped with a ZeissAxio Cam digital camera.
Preparation of purified synaptic plasma membrane vesicles from rat whole brain or hippocampus
Rats (Sprague Dawley, 150–250 gm) were killed by CO2 inhalation and decapitated; their brains were rapidly removed; the hippocampus was dissected; and the tissue placed in ice-cold homogenization media (320 mm sucrose, 10 mm Tris-HCl, pH 7.4, 1 mmK2-EDTA, 10 μm PMSF, 10 μm benzamidine, and 1 μm pepstatin A). Synaptic plasma membrane vesicles were prepared by fractionated centrifugation of homogenated brain tissue followed by 7.5/10.0% Ficoll gradient centrifugation (to isolate intact synaptosomes) (Lai et al., 1977). Synaptosomes were lysed, and the respective plasma membrane fraction was isolated by sucrose density gradient centrifugation (Vazquez et al., 1990). Plasma membrane enrichment through the different preparation steps was monitored using radioligand binding for established plasma membrane marker proteins (e.g., [3H]isradipine for L-type Ca2+ channels, [125I]ω-conotoxin GVIA for N-type Ca2+ channels, [125I]iberiotoxin for high-conductance Ca2+-activated K+ channels, and [125I]apamin for small-conductance Ca2+-activated K+ channels). On average, specific activity of the final membrane preparation was enriched 8- to 12-fold compared with the respective starting material (data not shown).
Preparation of Xenopus oocyte membranes expressing SK channels
For membrane preparation, batches of 60–80 oocytes injected with SK1, SK2, or SK3 (Kohler et al., 1996) or noninjected oocytes were homogenized in 10% sucrose dissolved in homogenization buffer [600 mm KCl, 5 mm3-(N-morpholino)propanesulfonic acid (MOPS), 100 μm PMSF, 1 μm pepstatin A, 1 μm p-aminobenzamide, 1 μg/ml aprotinin, and 1 μg/ml leupeptin, pH 6.8] in a volume of 10 μl/oocyte with 10 strokes at 1000 rpm in a glass–Teflon homogenizer (Braun-Melsungen) at 0°C. The homogenate was placed on top of a step gradient consisting of 7 ml of 50% sucrose and 3.5 ml of 20% sucrose in homogenization buffer and centrifuged at 67,000 × gfor 30 min at 4°C in a Beckman SW 40 rotor. The interface (between 20 and 50% sucrose) was collected and subjected to centrifugation at 84,000 × g for 30 min at 4°C in a Beckman Ti 70.1 rotor. The supernatant was discarded, and the pellet was resuspended in 200 μl of (in mm): 300 sucrose, 100 KCl, and 5 MOPS, pH 6.8, and stored at −80°C until use.
Solubilization of rat brain SK channels, immunoprecipitation studies, and in vitro [125I]apamin binding studies
Solubilization. Rat brain synaptic plasma membrane vesicles were sedimented (45,000 × g, 15 min), and the resulting pellet was resuspended in (in mm): 5 Tris-HCl, pH 7.4, 0.1 PMSF, 1 iodoacetamide, and 0.1 benzamidine containing 4% (w/v) sodium cholate and 500 mmKCl at a final protein concentration of 20 mg/ml. After incubation, with intermittent mixing, at 1°C for 60 min, insoluble material was sedimented at 106,000 × g for 60 min.
Immunoprecipitation and in vitro[125I]apamin binding studies. For all experiments, crude anti-SK1(12–29), anti-SK1(515–532), anti-SK2(538–555), and anti-SK3(504–522) serum or a combination thereof was prebound to an equal amount of packed protein A-Sepharose in radioimmunoassay (RIA) buffer (5 mm Tris-HCl, pH 7.4, 5 mm KCl, 0.1% BSA, and 0.3% sodium cholate) for 60–120 min under gentle rotation. The gel was washed three times with 1 ml of RIA buffer before the addition of sodium cholate-solubilized SK channels. The solubilized material was diluted twofold in RIA buffer to lower the detergent and KCl concentration and added to the prebound antibodies, and incubation was continued for 12 hr at 4°C. Each protein A-Sepharose pellet was split into six equal samples, and antibody-bound SK channels were determined by [125I]apamin binding (three samples for control binding and three samples for nonspecific binding). The incubation medium (200 μl) consisted of RIA buffer. Nonspecific binding was defined in the presence of 30 nm apamin, and incubation was performed at 4°C. After 60 min of incubation with the radioligand in the absence (control) or presence of apamin (nonspecific binding), the protein A-Sepharose was rapidly washed three times with ice-cold RIA buffer, and bound [125I]apamin was determined by gamma radiation counting. Under these conditions, a saturating concentration of anti-SK2(538–555) (e.g., 25 μl of serum) typically precipitated 11000–25000 cpm of [125I]apamin (<200 cpm in the presence of 30 nm apamin), whereas the respective preimmune serum precipitated <350 cpm of [125I]apamin (<200 cpm in the presence of 30 nm apamin).
Coimmunoprecipitation and cross-blotting.Anti-SK2(538–555) and anti-SK3(504–522) antibodies were affinity-purified as described previously (Knaus et al., 1995) and subsequently coupled to CNBr-activated Sepharose 4B according to the manufacturer's protocol. Solubilized material was diluted threefold in (in mm): 5 Tris-HCl, pH 7.4, and 150 NaCl buffer and incubated with the respective antibody column for 36 hr at 4°C. Columns were washed with 40 bed volumes of 0.1% Triton in (in mm): 5 Tris-HCl, pH 7.4, and 150 NaCl. Retained SK2 and SK3 channels were eluted with 0.1 m glycine, pH 2.5, 0.1% Triton X-100, and 150 mm NaCl. IgG leaching from the antibody column was diminished by addition of protein A-Sepharose to the eluted material for 1 hr at 20°C. Eluates were separated by SDS-PAGE and Western blotting was performed as described above.
Slice preparation. Young male Wistar rats (17–35 d) were deeply anesthetized with halothane before decapitation. Transverse hippocampal slices (400 μm thick) were prepared using a Vibratome(752M; Campden Instruments, Loughsborough, UK) and maintained in artificial CSF (ACSF) containing (in mm): 125 NaCl, 25 NaHCO3, 1.25 KCl, 1.25 KH2PO4, 1.5 MgCl2, 1 CaCl2, and 16 glucose and saturated with 95% O2 and 5% CO2.
Whole-cell recording and drug application. During recording, the slices were kept at room temperature and superfused with ACSF of the above composition, except that the concentration of CaCl2 was raised to 2 mm. Whole-cell gigaohm seal recordings were obtained from CA1 pyramidal cells using the “blind” method. The patch pipettes were filled with a solution containing (in mm): 140 K-gluconate, 10 HEPES, 2 ATP Na salt, 0.4 GTP Na salt, and 2 MgCl2, resulting in a pipette resistance of 4–7 MΩ. The cells were voltage-clamped using an Axopatch 1D amplifier (Axon Instruments, Foster City, CA), and signals were filtered at 2 kHz (−3 dB). The series resistance was 10–26 MΩ, and all potentials were corrected for the liquid junction potential (−10 mV). To record the SK channel-mediated K+ currents in relative isolation, tetrodotoxin (TTX, 1 μm) and tetraethylammonium (TEA, 5 mm) were routinely added to the extracellular medium to block Na+ channels and BK, M, and delayed rectifier K+ channels. SK channels are known to be relatively insensitive to TEA. Thus, Ishii et al. (1997)found that 5 mm TEA blocks only ∼25% of the current through SK1 homomeric channels expressed in Xenopusoocytes.
In some experiments, the selective M-channel blocker XE991 (10 μm) was added to the extracellular medium in addition to 5 mm TEA to block any remaining M-current. However, XE991 had no apparent additional effects, indicating that the M-current was already fully blocked by 5 mm TEA (Storm, 1989, 1990). The whole cell capacitance (C m) was calculated according to the formula C m= (∫dI/dt)/V; where V is the amplitude of a negative voltage-clamp step (−10 mV, 10 msec long) and ∫dI/dt is the time integral of the capacitative current transient after the end of the negative step.
Data acquisition, storage, and analysis
The data were acquired using pClamp 7.0 (Axon Instruments) at a sampling rate of 1 kHz, digitized (10 kHz), stored on videotapes (Instrutec VR-10), analyzed, and plotted using pClamp 7.0 and Origin 5.0 (Microcal). The peak of the mAHP current (I mAHP) was measured by averaging the amplitude measurements within a 10 msec time window 50 msec after the end of depolarizing step; this time window always corresponded to the peak of I mAHP. Because the time to peak was more variable for the sAHP current (I sAHP), we used a different method to measure its amplitude. The current trace within a time window that included I sAHP but excludedI mAHP was low-pass-filtered at 100 Hz to reduce the high-frequency noise, and the peak amplitude was determined at the time point at which the filteredI sAHP was maximal.
Values are expressed as mean ± SEM. Two-tailed paired Student'st test was used for statistical analysis (α = 0.05).
Characterization of SK-specific antibodies in immunoblot analysis of rat brain
Anti-SK1(12–29), anti-SK1(515–532), anti-SK2(538–555), and anti-SK3(504–522) sequence-directed antibodies against the pore-forming subunits of SK1, SK2, and SK3 channels were applied to investigate the presence and apparent molecular weight of their tissue-expressed gene products. All antibodies specifically recognized both the respective in vitro-translated SK channels (data not shown) and SK1, SK2, or SK3 protein after expression in Xenopus oocytes followed by isolation of the plasma membrane fraction (Fig. 1). InXenopus oocytes, the respective SK antibodies recognized immunoreactive bands of 69 kDa (SK1), 64 kDa (SK2), and 67 kDa (SK3).
In purified synaptic plasma membrane vesicles from rat whole brain, anti-SK1(12–29) stained three diffuse bands with apparent M r values of 65, 58, and 43 kDa, whereas the respective C-terminal antibody anti-SK1(515–532) exclusively stained polypeptides with apparent M r values of 65 and 58 kDa. An identical polypeptide pattern was observed for mouse whole-brain membranes (data not shown). In contrast, anti-SK2(538–555), and anti-SK3(504–522) stained single bands with overall M r values of 67 and 70 kDa, respectively (Fig. 1). The observed M rvalues for SK2 and SK3 channels are in good agreement with the deducedM r values (SK2, 64 kDa; SK3, 63 kDa; for considerations regarding the polypeptides detected by anti-SK1(12–29) and anti-SK1(515–532), see Discussion). The immunostaining signal was substantially reduced by inclusion of 10 μm immunogenic peptide (data not shown) and not present when using the respective preimmune sera (Fig. 1).
Subunit composition of apamin-sensitive SK channels in rat brain
To investigate further the molecular components of SK channels in rat brain, these channels were solubilized with sodium cholate; the individual channel population was immunoprecipitated by the corresponding antibodies; and the amount of channel precipitated was quantified by [125I]apamin binding. In all cases, the respective preimmune sera precipitated <5% compared with the corresponding immune sera (Fig.2 B). Although anti-SK2(538–555) and anti-SK3(504–522) yielded saturable levels of [125I]apamin-binding precipitation, both anti-SK1(12–29) and anti-SK1(515–532) failed to immunoprecipitate [125I]apamin binding to a significant extent (Fig. 2 A). Anti-SK2(538–555) precipitated 71 ± 5% (n = 3) of soluble apamin receptors, whereas anti-SK3(504–522) precipitated 29 ± 7% (n = 3) of the sites. However, the combination of both antisera resulted in strict additivity, indicating complete segregation of SK2 and SK3 channels into individual channel populations (Fig.2 B). Western blot analysis of the remaining supernatant indicated complete clearance from soluble apamin-sensitive SK channels (data not shown). Additional support stems from cross-blotting experiments using anti-SK2(538–555) and anti-SK3(504–522) immunoaffinity columns (Fig.2 C). Although anti-SK2(538–555) is clearly capable of retaining SK2 channels, this antibody failed to coimmunoprecipitate SK3 channels. In support of these data, anti-SK3(504–522) immunoaffinity purified solely SK3 channels; however, no SK2 channels were detected through this cross-blotting approach.
Because [125I]apamin failed to detect SK1 channels being precipitated by the respective antibodies, this cross-blotting approach was also attempted. However, neither of the anti-SK antibodies was capable of immunoprecipitating detergent-solubilized SK1 protein. This finding indicates that none of our anti-SK1 antibodies is suitable for immunoprecipitation experiments; therefore, no conclusion regarding subunit composition or association of SK1 channels with calmodulin can be drawn.
After providing evidence that SK2 and SK3 subunits do not coassemble to form heterotetrameric SK2/SK3 channels, we next investigated whether calmodulin is constitutively bound to rat brain SK channels. As shown in Figure 2 D, both anti-SK2(538–555) and anti-SK3(504–522) coimmunoprecipitated calmodulin. This precipitation was specific, because a corresponding amount of the respective preimmune serum failed to precipitate calmodulin. The result with the anti-SK3 antibody is in good agreement with a previous report (Xia et al., 1998).
Regional distribution of SK protein in neocortex and hippocampal formation
Specificity of immunostaining
The neuronal distribution pattern of all three subunits was investigated by immunohistochemical analysis (Figs.3, 4). The characteristic immunostaining pattern fulfilled the specificity criteria, because it was not observed with the corresponding preimmune sera (Fig. 3) or in the presence of excess immunogenic peptide (data not shown). It is beyond the scope of this paper to provide a detailed description of the distribution of SK1-, SK2-, and SK3-IR in all brain regions. However, the salient features of the immunohistochemical staining patterns of these three proteins are described below.
Expression of SK1 protein
To obtain a distribution profile of all SK1 proteins independently of alternate splicing, the anti-SK1(12–29)antibody was used for all distribution studies. In the neocortex, almost all SK1 IR was restricted to fibers extending from layer V to layer I (Fig. 4 A). The staining appeared to be associated with both the proximal and distal dendrites of the layer V pyramidal cells. Little or no IR could be observed in perikarya in this brain region.
However, the highest levels of SK1 protein were expressed in the hippocampal formation, again showing staining associated with the neuropil (Fig. 4 D,G,J). Virtually no immunostaining was associated with granule and pyramidal cell somata. The molecular layer (inner, middle, and outer parts) of the dentate gyrus, containing the dendrites of granule, basket, and polymorphic cells as well as perforant path terminals, revealed the most prominent staining (Fig. 4 J). Dense IR could also be observed within the mossy fiber system, with highest levels in the stratum lucidum of the CA3 area. This layer contains three principal structures: pyramidal cell dendrites, mossy fiber axons, and synaptic complexes. Somewhat lower concentrations of the SK1 protein could be detected in stratum oriens and stratum radiatum (Fig.4 D). These regions contain, for example, the dendrites of the pyramidal cells as well as the Schaffer collaterals. Interestingly, the signal was slightly discontinuous at the junction between the CA1 and CA3 regions, with a somewhat higher density of IR in CA1 (Fig. 4 G).
Expression of SK2 protein
SK2-IR in the neocortex could be predominantly observed in pyramidal cells of layer V. Distinct staining was associated with the soma as well as with the proximal portion of the dendritic tree (Fig.4 B). All other neocortical compartments appeared to be almost devoid of SK2-IR.
In the hippocampal formation, SK2-IR was mainly enriched in the CA1–CA3 region (hippocampus proper) but predominately spared the dentate gyrus (Fig. 4 H). Within the hippocampus proper, the highest level of SK2 protein was present in the stratum oriens (especially in the layer of the proximal dendrites of pyramidal cells), whereas the strata radiatum and pyramidale showed intermediate to low levels of immunoreactivity, respectively. No SK2 immunostaining was associated with the stratum lucidum of CA3 (Fig.4 E). As mentioned before, the dentate gyrus (DG) revealed only low levels of SK2 protein. Faint staining could be observed in the molecular layer, whereas the DG granule cell layer was virtually devoid of SK2-IR (Fig. 4 K).
Additionally, a limited number of hippocampal interneurons were found to expose SK2-IR. Most of these were found in the stratum oriens, although scattered neurons could be observed throughout the hippocampus.
Expression of SK3 protein
In contrast to SK1 and SK2 protein, the anti-SK3(504–522) antibody revealed only faint staining throughout the entire neocortex. No particular enrichment in any layer could be observed. The most prominent feature was the staining of varicose fibers in all neocortical layers (Fig.4 C).
The hippocampal formation showed moderate levels of SK3-IR compared with its expression in the basal ganglia or thalamus. SK3-IR was almost uniformly distributed throughout this brain region but with distinct fiber staining within the hilus and the terminal field of mossy fibers (Fig. 4 I,L). This suggests an axonal or presynaptic localization of SK3 protein for areas that were mostly devoid of SK2 protein. Virtually no IR could be found for the DG granule cell layer and the superficial molecular layer (Fig. 4 L). Moderate levels of SK3-IR were also associated with the stratum lacunosum moleculare, a layer in which the perforant path fibers from the entorhinal cortex terminate. A low level of immunostaining was present in the strata oriens and radiatum of CA1–CA3, whereas no IR could be detected in the pyramidal cell layer (Fig.4 F,I).
Taken together, the staining patterns of SK1 and SK2 protein were found to be distinct; however, they overlapped in some brain regions. SK3 protein was distributed mostly reciprocally without any significant extent of overlap to SK1 and SK2.
Functional expression of SK channels in rat brain
Given the prominent expression of SK1–SK3 protein in the rat hippocampal formation, we focused our electrophysiological experiments on this brain region.
Whole-cell somatic recordings were obtained from CA1 and CA3 pyramidal cells and granule cells of the dentate gyrus in rat hippocampal slices. To compare the Ca2+-activated K+ currents (AHP currents; Pennefather et al., 1985; Lancaster and Adams, 1986; Storm, 1990; Stocker et al., 1999) in the three cell populations, each cell was voltage-clamped at a holding potential of −50 or −55 mV. One micromolar TTX and 5 mm TEA were routinely added to the extracellular medium to record the SK currents in relative isolation (see Materials and Methods).
Calcium-activated tail currents were elicited by a brief (100 msec) depolarizing voltage step, which reliably triggered Ca2+ influx in the form of an unclamped Ca2+ action current (see Materials and Methods; Pedarzani and Storm, 1993). In the CA1 and CA3 pyramidal cells, each depolarizing step was followed by a biphasic outward tail current (AHP currents) consisting of early and late components (Fig.5 A). The early outward tail current contributes to the mAHP (Storm, 1989; Stocker et al., 1999), whereas the slow component, lasting 4–7 sec, is called theI sAHP because it underlies the slow AHP (Hotson and Prince, 1980; Lancaster and Adams, 1986; Storm, 1990;Sah, 1996). Both I sAHP and the apamin-sensitive component of the mAHP current were largely resistant to 5 mm TEA but were readily suppressed by perfusion with Ca2+-free medium and by the Ca2+ channel blockers Mn2+ and Cd2+(data not shown; Lancaster and Adams, 1986; Pedarzani and Storm, 1993;Stocker et al., 1999).
Figure 5 shows typical AHP currents recorded in a CA1 pyramidal cell (Fig. 5 A), a CA3 pyramidal cell (Fig. 5 B), and a DG granule cell (Fig. 5 C). The CA1 and CA3 pyramidal cells showed a large mAHP current followed by a smaller sAHP current. In contrast, the DG cells showed little or no distinct early (mAHP) current, although the I sAHP amplitude was similar to that of CA1 and CA3 pyramidal cells (Fig. 5C). Furthermore, whereas bath application of 100 nmapamin blocked the large mAHP current in both the CA1 and CA3 pyramidal cells, this toxin had only a small effect on the early tail currents of the DG granule cells (Fig. 5 C). This indicates that the apamin-sensitive SK current is substantially smaller in the latter cell type. In contrast, the sAHP current was relatively similar both in amplitude and time course in the three cell types (Fig. 5, Table1). Apamin (10 μm) had no measurable effect onI sAHP in any of the three cell types (Fig. 5 A–C), indicating that this current is generated by apamin-resistant channels (Lancaster and Nicoll, 1987; Storm, 1989).
In the CA3 cells, apamin unmasked an inward tail current (Fig.5 B). The nature of this current was not determined in our study, but it is likely to be a Ca2+ or Na+ current, possibly through Ca2+-activated nonselective cation channels (Partridge and Swandulla, 1993). A similar inward tail current that was observed in the DG cells after blocking other K+ currents was constant throughout the recordings and apparently not affected by apamin (see below). It therefore seems unlikely that our I aAHPmeasurements were significantly confounded by the inward tail current.
To study the apamin-sensitive current, I aAHP, in relative isolation, we performed the remaining experiments under conditions that suppressed I sAHP as well as other K+ currents (Fig.6). In these experiments, M-, BK-, and delayed rectifier-type K+ currents were suppressed by bath application of 10 μm XE991 and 5 mm TEA. In addition,I sAHP was suppressed by inclusion of cAMP analogues [in μm: 100 cAMP or 100 8 chlorophenylthio (CPT)-cAMP] in the intracellular medium in the recording pipette (Madison and Nicoll, 1986). The SK channels are known to be predominately resistant to these drug concentrations (Ishii et al., 1997; Wang et al., 1998; Hu et al., 2001), but evidence for a small cAMP-induced increase of I aAHPhas been reported (Stocker and Pedarzani, 2000). The blockade of non-SK K+ channels also served to increase the cell input resistance, thereby facilitating the activation of voltage-gated Ca2+ channels and, hence, Ca2+ influx and SK channel activation and improving the space clamp of the cell. Figure 6 shows representative AHP currents from CA1 (Fig. 6 A), CA3 (Fig.6 B), and DG (Fig. 6 C) neurons under these conditions.
In response to the same voltage-clamp protocol as used in Figure 5, the CA1 and CA3 pyramidal cells still showed an outward mAHP current, whereas the DG cells showed only an inward tail current under these conditions. Application of apamin again suppressed virtually all mAHP current in CA1 (n = 9) and CA3 (n = 8) pyramidal cells but had only a small effect in the DG granule cells. Furthermore, 100–600 μm d-tubocurare applied in addition to 100 nm apamin produced no further effect (n = 20), suggesting that 100 nmis a saturating concentration of apamin under these conditions (Kohler et al., 1996; Stocker et al., 1999). To isolate the apamin-sensitive current, records taken after the toxin application were digitally subtracted from the control records obtained just before the application (Fig. 6 A–C, right traces). The amplitude of the difference current,I aAHP, was 130–150 pA in the CA1 and CA3 pyramidal cells but only ∼30 pA in the DG granule cells (Fig. 6 D,E). In all cases, the difference inI aAHP amplitude between granule and pyramidal cells was highly significant (p = 0.00006–0.007; see Table 3).
The apamin-sensitive current was somewhat larger and more slowly decaying in the CA3 than in the CA1 pyramidal cells (Figs. 5,6 A,B). Thus, for cells recorded with cAMP, the decay time constant ofI aAHPwas 177 ± 30 msec in the CA1 (n = 9) versus 297 ± 23 msec in CA3 (n = 8; p = 0.008; Tables 2, 3). A similar difference was also found in cells recorded without cAMP (Table 1). In addition toI aAHP, two of the eight CA3 cells recorded with cAMP also showed a slower apamin-insensitive outward current component, which was not characterized further in this study.
The small size of the apamin-sensitive SK current in the DG granule cells was apparently not attributable to failure of Ca2+ influx during the depolarizing step, because DG granule cells showed a clear Ca2+ spike during each voltage step (data not shown), as well as a prominentI sAHP, which was also dependent on Ca2+ influx (Fig. 5 C, Table 1) (Haas and Rose, 1987). We observed no apamin-induced changes in the depolarization-induced Ca2+ currents that triggered the AHP currents in each of the three cell types (data not shown), indicating that the action of apamin is downstream of the Ca2+ influx.
The difference in whole-cell current is probably related to differences in cell size. To relate the current measurements to the surface membrane area, we calculated the membrane capacitance for each cell from capacitative transients evoked by hyperpolarizing voltage steps (see Materials and Methods) and divided the currents by the capacitance. For the I aAHP, the current densities thus obtained in the CA1, CA3, and DG cells were 0.80, 0.75, and 0.38 pA/pF, respectively. Thus, theI aAHP density in the DG granule cells appears to be only approximately half of that in the CA pyramidal cells (Fig. 6 G). In contrast, the average density of the sAHP current was somewhat larger in the DG granule cells (0.45 ± 0.11 pA/pF) than in the CA1 and CA3 pyramidal cells (0.29 ± 0.08 and 0.16 ± 0.04 pA/pF, respectively), although the difference was not statistically significant in our data (Fig. 5 G, Tables1-3).
In conclusion, there seems to be a substantial difference in the size of the apamin-sensitive SK current between the CA1 and CA3 hippocampal pyramidal cells on one hand and the DG granule cells on the other, whereas the apamin-resistant sAHP current shows no such difference. This holds both for the total current per cell and when corrected for cell size (membrane capacitance, reflecting surface area).
This report presents the first distribution profile for all SK channel subunits in a mammalian brain. The data indicate distinct, although partly overlapping, distributions for the three SK subunits and suggest that the apamin-sensitive medium afterhyperpolarization of the hippocampal pyramidal cells is mainly attributable to SK2 protein.
To obtain a complete antibody panel for the individual SK channel subunits, we raised a total of 14 sequence-directed anti-SK channel antibodies of which four antibodies were used to perform this study. The difficulties obtaining suitable SK channel antibodies with high selectivity were mostly attributable to the very low expression density of SK channel protein in mammalian brain (based on radioligand binding data using the SK channel ligand [125I]apamin). This toxin labels only 20–40 fmol of binding sites/mg of protein, a density 10 times lower than for BK-channels (as measured by [125I]iberiotoxin binding; Koschak et al., 1997) or only 1% of the density of toxin-sensitive voltage-gated K+ channels (e.g., measured by [125I]hongotoxin binding; Koschak et al., 1998) in the same membrane preparation.
Additional complications stem from the fact that the rodent SK1 gene (but not the SK2 and SK3 genes) undergoes extensive alternative splicing. Mouse and rat brain express at least eight 3′-variant SK1 transcripts with additional heterogeneity in the 5′ region (Shmukler et al., 2001). To obtain a clear picture of the distribution of the entire SK1 channel population, N- and C-terminal SK1 antibodies were used. In immunoblotting experiments the N-terminal antibody revealed the tissue expression of three SK1 polypeptide families with apparentM r values of 65, 58, and 43 kDa, respectively (Fig. 1). The reported cDNA sequences also subdivided the putative SK1 polypeptides into three distinct groups with predictedM r values of 62, 57–58, and 45–49 kDa. A C-terminal anti-SK1 antibody detected only polypeptides with apparent M r values of 65 and 58 kDa, respectively. These results are in agreement with the alternative splicing data (the truncated SK1 polypeptides are expected to lack the respective antibody recognition sequences). Remarkably, we did not obtain any evidence for alternative splicing at the N terminus. In contrast to SK1 protein, the anti-SK2 and anti-SK3 antibodies labeled only single polypeptides with apparentM r values of 67 and 70 kDa, respectively.
After having established that our antibodies specifically recognize their respective target proteins, we used them to establish the distribution of all known SK channels. The three SK channel subunits displayed distinct although sometimes overlapping distribution. Most of the higher brain regions such as the neocortex and hippocampus showed expression of both SK1 and SK2 channels, whereas phylogenetically older brain regions (e.g., the thalamus, basal ganglia, cerebellum, and brainstem) showed high levels of SK3 expression. The strong expression of SK3 channel protein in the basal ganglia and brainstem was also previously observed for the mouse brain (Bond et al., 2000).
A particular informative distribution pattern was observed for the hippocampal formation. High levels of SK2 expression in the CA1–CA3 fields contrasted with the low levels in the dentate gyrus. Conversely, for SK1 protein, high levels were detected in the DG with lower levels in CA1–CA3. Immunoreactivity for these SK proteins was found in regions containing densely packed dendrites of pyramidal cells (SK1- and SK2-IR) or DG granule cells (SK1-IR). Considering that the granule cells are more numerous and smaller, with thinner dendritic shafts, than the pyramidal cells (Amaral et al., 1990), there may be more neuronal plasma membrane area per volume unit in the DG than in CA1–CA3. If so, the relatively weak SK2 staining in the DG may reflect a still lower density of SK2 protein per membrane area relative to CA1–CA3. Conversely, the stronger SK1 staining in the DG relative to CA1–CA3 may not necessarily represent a higher density of SK1 protein per membrane area.
SK3 protein in rat hippocampus is expressed only at moderate to low levels (Figs. 3, 4) with the most prominent staining observed in the mossy fiber system. This distribution in conjunction with the characteristic immunostaining of varicose fibers throughout the hippocampal formation suggests most likely an axonal or presynaptic localization of SK3 protein. This proposed subcellular targeting is supported by recent colocalization studies in cultured neonatal hippocampal neurons in which SK3 protein clearly coresides with established presynaptic marker proteins (e.g., synapsin; Obermair et al., 2001).
On the basis of the distinct distribution pattern in the hippocampal formation, this brain region was chosen for functional investigations. Whole-cell patch-clamp recordings revealed a striking difference in the functional expression of the apamin-sensitive outward currentI aAHP between the DG granule cells and the CA1–CA3 pyramidal cells. Whereas the pyramidal cells showed a robust I aAHP of ∼100 pA (80–130 pA in CA1 and 120–150 pA in CA3), I aAHPwas an order of magnitude smaller in the granule cells (∼8 pA under normal conditions) (Figs. 5, 6, Table 1). Even when corrected for the membrane capacitance of the cells (which is approximately proportional to their surface area), the currents (i.e., current densities) were still approximately twice as high in CA1–CA3 as in DG (∼0.8 pA/pF in CA1–CA3 vs 0.4 pA/pF in DG). In contrast, the average density of the slow AHP current was somewhat larger in the DG than in CA1–CA3.
The apamin-sensitive SK current (∼30 pA) detected in DG granule cells was significantly enhanced when the cells were perfused with cAMP analogues and other outward currents were suppressed (p = 0.002). This difference and the increased mean I aAHP in the pyramidal cells treated with cAMP may reflect a cAMP-induced upregulation of SK channels (as suggested for CA1 cells; Stocker et al., 1999). However, the difference in the observed I aAHPamplitude between CA1–CA3 and the DG cannot be explained by different cAMP regulation of the SK channels in the two regions, because the contrast was observed either with or without cAMP in the pipette. In addition to cAMP effects, improved clamp conditions may contribute to the larger SK currents observed during blockade of other K+ currents. The difference in SK current is probably not attributable to a difference in Ca2+ influx between CA1–CA3 and DG cells, because the Ca2+ currents were stable throughout the recordings and elicited robust sAHP currents of similar amplitude in all three cell types (Figs. 5, 6, Tables 1-3).
There seems to be a close agreement between the functional expression of apamin-sensitive current and the distribution of SK2 protein in the hippocampal principal neurons. Thus, the SK2 protein levels, as well as the apamin-sensitive current amplitudes, were highest in CA1–CA3 and substantially lower in the DG. In contrast, the differences in currents do not match the SK3 protein distribution, which was rather uniform throughout the entire hippocampal formation. This suggests that SK2 underlies the apamin-sensitive mAHP component of the CA1 and CA3 pyramidal cells. Because only SK2 homomultimers were detected by coimmunoprecipitation (Fig. 2 B,C), it is likely that this channel type underlies the apamin-sensitive mAHP and, hence, the apamin-sensitive early spike frequency adaptation in the hippocampal pyramidal cells. However, it remains to be determined whether SK1 protein contributes to the apamin-sensitive AHP in the DG granule cells, where this SK species is more strongly expressed.
Why does the decay time course of the apamin-sensitive current differ between CA1 and CA3 pyramidal cells? This seems to most simply be explained by different Ca2+ dynamics, subcellular distribution of SK2 channels, or both, between the two cell types. The time course of the SK2 and SK3 channel activity is known to closely follow the rise in intracellular Ca2+ (Vergara et al., 1998). The SK channels open and close rapidly in response to changes in intracellular [Ca2+] and show little or no intrinsic time dependence at the time scale relevant for the overall tail current kinetics. Thus, the SK current time course probably reflects the Ca2+ dynamics of the particular cell type and subcellular domain where the channels are located, independently of the SK channel subtype involved. Therefore, the difference in time course between CA1 and CA3 seems fully compatible with the hypothesis that SK2 homomultimers generate the apamin-sensitive mAHP in both cell types.
The relatively large sAHP current density in DG granule cells compared with CA1–CA3 pyramidal cells seems to approximately parallel the high levels of SK1 protein in the DG relative to CA1–CA3. This, along with the lower apamin sensitivity of homomeric SK channels, may seem to support the hypothesis that SK1 underlies the apamin-insensitive sAHP. However, this idea seems still difficult to reconcile with the observation that the hippocampal sAHP seems to be entirely resistant to high concentrations of apamin, which completely block the homomultimeric SK1 channels tested in different expression systems (Shah and Haylett, 2000a; Strobaek et al., 2000; Grunnet et al., 2001b). This hypothesis might, however, be rescued if there exist auxiliary (β) subunits that substantially alter the toxin sensitivity of the SK channel complex, such as the BK-β4 subunits have been found to do for BK channels (Meera et al., 2000).
This work was supported by European Community Grants BMH4-CT97-2118 and QLRT-1999-01356 (J.F.S. and H.-G.K), Austrian Research Foundation Grant P14954-PHA (H.-G.K), the Norwegian Medical Research Council, and the Nansen and Odd Fellow Foundations. This report is part of the PhD theses of C.A.S. and H.H. We thank Dr. Hartmut Glossmann for continuous support. Xenopus oocyte membranes were kindly supplied by Dr. Dan Klaerke, Pannum Institute (Copenhagen, Denmark).
Correspondence should be addressed to Hans-Günther Knaus, Institut für Biochemische Pharmakologie, Peter-Mayr Strasse 1, A-6020 Innsbruck, Austria. E-mail:.