Deletion of the Kv2.1 delayed rectifier potassium channel leads to neuronal and behavioral hyperexcitability

Abstract

The Kv2.1 delayed rectifier potassium channel exhibits high-level expression in both principal and inhibitory neurons throughout the central nervous system, including prominent expression in hippocampal neurons. Studies of in vitro preparations suggest that Kv2.1 is a key yet conditional regulator of intrinsic neuronal excitability, mediated by changes in Kv2.1 expression, localization and function via activity-dependent regulation of Kv2.1 phosphorylation. Here we identify neurological and behavioral deficits in mutant (Kv2.1^−/−) mice lacking this channel. Kv2.1^−/− mice have grossly normal characteristics. No impairment in vision or motor coordination was apparent, although Kv2.1^−/− mice exhibit reduced body weight. The anatomic structure and expression of related Kv channels in the brains of Kv2.1^−/− mice appears unchanged. Delayed rectifier potassium current is diminished in hippocampal neurons cultured from Kv2.1^−/− animals. Field recordings from hippocampal slices of Kv2.1^−/− mice reveal hyperexcitability in response to the convulsant bicuculline, and epileptiform activity in response to stimulation. In Kv2.1^−/− mice, long-term potentiation at the Schaffer collateral – CA1 synapse is decreased. Kv2.1^−/− mice are strikingly hyperactive, and exhibit defects in spatial learning, failing to improve performance in a Morris Water Maze task. Kv2.1^−/− mice are hypersensitive to the effects of the convulsants flurothyl and pilocarpine, consistent with a role for Kv2.1 as a conditional suppressor of neuronal activity. Although not prone to spontaneous seizures, Kv2.1^−/− mice exhibit accelerated seizure progression. Together, these findings suggest homeostatic suppression of elevated neuronal activity by Kv2.1 plays a central role in regulating neuronal network function.

Introduction

Voltage-dependent K⁺ (Kv) channels are important regulators of neuronal excitability and represent attractive candidates for therapeutic modulation of neuronal activity, including neuronal hyperexcitability that occurs in epileptic and stroke patients (Castle, 2010, Judge et al., 2007, Meldrum & Rogawski, 2007, Wulff et al., 2009). Delayed rectifier Kv currents (I_K) mediated by the Kv2 family are important in regulating somatodendritic excitability in a wide variety of mammalian neurons, including hippocampal and cortical pyramidal neurons (Bekkers, 2000, Du et al., 2000, Guan et al., 2007, Korngreen & Sakmann, 2000, Mohapatra et al., 2009). Inhibition of Kv2.1 channels using oligonucleotide-mediated knockdown, or acutely with neurotoxins, or blocking antibodies suggests that currents mediated via Kv2.1 channels constitute the major component of somatodendritic I_K (Du et al., 2000, Guan et al., 2007, Malin & Nerbonne, 2000, Martina & Jonas, 1997, Mohapatra et al., 2009, Murakoshi & Trimmer, 1999, Pal et al., 2003). In brain neurons Kv2.1 is phosphorylated at a large number of sites located on its extensive cytoplasmic domains, and this phosphorylation exhibits dramatic bidirectional activity-dependent regulation (Cerda & Trimmer, 2011, Misonou et al., 2006b, Park et al., 2006). Increased neuronal activity induced by seizures and hypoxia in vivo, or glutamate stimulation and chemical ischemia in cultured hippocampal neurons, leads to dephosphorylation of Kv2.1 through a Ca²⁺/calcineurin-dependent mechanism (Cerda & Trimmer, 2011, Misonou et al., 2006b, Misonou et al., 2005, Misonou et al., 2004, Misonou et al., 2008). In cultured neurons, these treatments lead to increased Kv2.1 activity due to large (≈30 mV) hyperpolarizing shifts in voltage-dependent activation (Mandikian et al., 2011, Misonou et al., 2006b, Misonou et al., 2005, Misonou et al., 2004, Mohapatra et al., 2009, Mulholland et al., 2008). Reducing neuronal activity by activity blockade in vitro or with anesthetics in vivo leads to hyperphosphorylation of Kv2.1, suggesting that regulation of Kv2.1 is bidirectional (Misonou et al., 2006b). Interestingly, studies of somatodendritic I_K in pyramidal neurons suggest that this current plays a major role in regulating excitability and Ca²⁺ influx during periods of repetitive high-frequency firing (Bekkers, 2000, Colbert & Pan, 2002, Du et al., 2000, Kang et al., 2000, Korngreen & Sakmann, 2000). In this context, the activity-dependent regulation of Kv2.1 channels allows Kv2.1 to act as a homeostatic suppressor of neuronal activity (Ikematsu et al., 2011, Misonou et al., 2005, Mohapatra et al., 2009, Surmeier & Foehring, 2004). The widespread expression of Kv2.1 suggests it could play a major role in homeostatic suppression of electrical excitability throughout the brain.

Specific roles have been ascribed to certain Kv channels through the use of subtype-specific blockers, for example a specific role for Kv1.1 channels in mediating neuroligin 1-dependent changes in interneuron excitability was recently elucidated in studies employing the Kv1.1-specific blocker dendrotoxin-K (Li et al., 2012). In contrast, the in vivo roles of Kv2.1 have been more difficult to assess due to the lack of selective pharmacological agents to specifically modulate Kv2.1 function. Kv2.1^−/− mice have been previously characterized for the impact of Kv2.1 ablation on glucose-stimulated electrical activity in pancreatic β cells (Jacobson et al., 2007). Here we utilize Kv2.1^−/− mice to perform the first comprehensive characterization of the behavioral and neurological phenotypes of mice in which Kv2.1 is ablated. Because Kv2.1 is expressed highly in the hippocampus, where it is known to suppress neuronal hyperexcitability, we chose to focus on behaviors influenced by hippocampal function, including seizure susceptibility and spatial learning. Our studies provide the first in vivo evidence of the crucial role for Kv2.1 in controlling neural function and behavior.

Materials and Methods

Animals

All animal use was approved by the Institutional Animal Care and Use Committee of the University of California, Davis, in accordance with the guidelines for animal use laid out by the US Public Health Service. Unless stated otherwise, all mice were group-housed in a climate-controlled facility on a 12-hour light/dark cycle (lights on at 06:00), and food (Teklad 2918, Harlan) and water were available ad libitum. Corncob bedding with a handful of shredded paper (Eco-Bedding^™, FiberCore, Cleveland, OH) lined the cage bottom. Unless specified otherwise all testing was performed during the light phase.

A full description of the generation of the Kv2.1 knockout line (official allele designation: Kcnb1^tmDgen) used for these studies can be found in a previous publication (Jacobson et al., 2007). Briefly, a targeting construct was designed to disrupt the Kv2.1 coding region after codon 346. Correctly targeted ES cells derived from the 129 strain were used to generate chimeric mice which were backcrossed to the C57BL/6J strain for about 10 generations. Upon receipt, we confirmed the genetic background of the Kv2.1^−/− mice to be >97% C57BL/6J using a panel of 1449 SNPs (Illumina, San Diego, CA). The introgressed region surrounding the Kv2.1 mutation was less than 10 Mb (Fig. S1). Genotyping assays were performed at DartMouse^™ (Dartmouth, NH). Heterozygous carriers (Kv2.1^+/−) were intercrossed to generate Kv2.1^+/+ and Kv2.1^−/− littermates for behavioral analysis. Unless noted otherwise, only male mice were used for experiments. We used an ear tagging method of identification. Because we noted significantly lower weight in Kv2.1 mutants older than about 3 months, unless noted otherwise, all behavioral testing was performed between 8 and 13 weeks of age.

Immunoblots

Animals were killed by rapid decapitation without anesthesia. Mouse brains were dissected and used to prepare mouse brain membrane (MBM) fractions as previously described (Misonou et al., 2006a). MBM fractions (40 μg protein/lane) were separated on a 7.5% SDS-PAGE gel and transferred to a nitrocellulose membrane. Immunolots were blocked in 4% non-fat dry milk/0.1% Tween-20/TBS (TBS: 20mM Tris/150 mM NaCl pH 8.0), and probed for Kv2.1 (K89/34), Kv2.2 (K37/89), Kv1.2 (K14/16), Kv1.4 (K13/31), Kv1.5 (K7/45) and Mortalin/Grp75 (N52A/42) using monoclonal antibodies (clone number in parentheses) obtained from the UC Davis/NIH NeuroMab Facility. Alexa-conjugated fluorescent secondary antibodies (Invitrogen, San Diego, CA) were used to detect bound primary antibodies using imaging with a FluorChem Q imager (Cell Biosciences). Immunoblots were stripped (stripping buffer: 2% SDS, 100 mM β-mercaptoethanol, and 62.5 mM Tris-HCl, pH 6.8) and re-probed so that each MBM sample was analyzed for expression of all of the antigens listed. Immunoblots were analyzed with the integrated density function in ImageJ (NIH) and statistical analysis was done using a Student’s t-test.

Immunohistochemistry

Animals were anesthetized with pentobarbital and transcardially perfused with 4% formaldehyde freshly prepared from paraformaldehyde. Sagittal brain sections (40 μm thick) were prepared from perfusion fixed animal and stained using free-floating methods as previously described (Rhodes et al., 2004). Briefly, sections were blocked and permeabilized with 10% goat serum, 0.3% TritonX-100 in 0.1 M phosphate buffer and incubated overnight in primary antibodies [Kv2.1 (KC and K89/34), Kv1.4 (K13/31), and Kv2.2 (N372B/60)]. Sections were exposed to Alexa-conjugated fluorescent secondary antibodies (Invitrogen, San Diego, CA) and Hoechst stain for 1 hour. Images of sections from Kv2.1^+/+ and Kv2.1^−/− mouse brains were taken using the same exposure time to compare the signal intensity directly, using a AxioCam HRm high-resolution CCD camera installed on an Axiovert 200M or AxioObserver Z1 microscope with 63x, 1.3 numerical aperture (NA) lens or 20x, 0.8 NA lens, and an ApoTome coupled to Axiovision software (Zeiss, Oberkochen, Germany). Images were identically processed in Photoshop to maintain consistency between samples.

Hippocampal culture preparation and whole cell voltage clamp recordings

Separate primary hippocampal neuronal cultures from each of the individual pups in a litter from heterozygous breeders were prepared as described (Beaudoin et al., 2012, Chen et al., 2008). Hippocampi were dissected from decapitated P0 mouse pups and put into 1.5 ml microcentrifuge tubes tubes, washed with Hank’s Balanced Salt Solution (HBSS) (Invitrogen #) and treated with 0.03% trypsin (Sigma #) in HBSS for 15 min at 37 °C. After two washes with HBSS, trypsin was inactivated with neuronal medium (Neurobasal medium (Invitrogen) supplemented with NS21 (Chen et al., 2008), 0.5 mM glutamine, 10 mM HEPES) plus 5% Horse Serum (HS) (Invitrogen #26050-088), and the tissue was triturated with a fire-polished Pasteur pipette. After allowing non-dissociated tissue pieces to settle, cells in the supernatants were counted and plated on coverslips (Warner Instruments #CS-12R) coated with 0.1% (w/v) poly-L-lysine (Peptides International #OKK-3056) at a cell density of 2 × 10⁵ cells/cm² in neuronal medium plus 5% HS. After 4 h, the medium was replaced with serum-free neuronal medium. Cells were grown at 37 °C in a humidified environment of 6.5% CO₂ for 4 days and 5% CO₂ thereafter.

Whole cell voltage clamp recordings were used to measure currents from dissociated hippocampal pyramidal neurons from Kv2.1^+/+ and Kv2.1^−/− mouse hippocampi after 4–6 days in culture (i.e., 4–6 DIV). Experimenters were blinded to the genotype of the mice during the experiments. The external (bath) solution contained (in mM): 3.5 KCl, 155 NaCl, 1.5 CaCl₂, 1 MgCl₂, 0.01 CdCl₂, 10 Glucose, and 10 HEPES, adjusted to pH 7.4 with NaOH. The internal (pipet) solution contained (in mM): 140 KCl, 13.5 NaCl, 1.8 MgCl₂, 0.09 EGTA, 4 Na-ATP, 0.3 Na-GTP, and 9 HEPES, adjusted to pH 7.2 with KOH. A calculated liquid junction potential of 4.1 mV was corrected. Pipette tips were coated with Sylgard 184 (Dow Corning #2010518) and fire polished. Pipette tip resistances with these solutions were less than 3 MΩ. Recordings were at room temperature (22–24° C). Voltage clamp was achieved with an Axon Axopatch 200B amplifier (MDS Analytical Technologies) run by Patchmaster software (HEKA). Holding potential was −70 mV to inactivate the majority of A-type current. Series resistance compensation was used when needed to constrain calculated voltage error to less than 10 mV in the cell body. When indicated, capacitance and Ohmic leak were subtracted using a P/5 protocol. Recordings were low pass filtered at 10 kHz and digitized at 100 kHz. All recordings were made after addition of vehicle: 5 μM tetrodotoxin (Abcam Biochemicals #ab120054), 100 nM Margatoxin (Peptides International # PAR-4290-s), and 0.1% BSA (Roche #03117332001). GxTX was a custom synthesized, oxidation resistant variant of the tarantula peptide guangxitoxin-1E (Herrington et al., 2006), with methionine 35 replaced by norleucine. Toxins were added by flushing 200 μl through a low volume recording chamber (Warner Instruments #R-24N). Voltage clamp data was analyzed and plotted with IgorPro software (Wavemetrics) as described previously (Sack et al., 2004).

Hippocampal slice preparation and electrophysiology

Mice were decapitated, and the brains were rapidly removed and placed in ice-cold artificial cerebrospinal fluid (ACSF) of the following composition (in mM): 124 NaCl, 5 (or 3 mM) KCl, 1.25 NaH₂PO₄, 2 MgSO₄, 2 CaCl₂, 26 NaHCO₃, and 10 glucose, which was saturated with 95% O2/5% CO2. Transverse hippocampal slices were made using a vibratome (Leica VT 1000A) at a thickness of 350 μm. The slices were transferred to a holding chamber containing continuously oxygenated ACSF and incubated to recover for at least 1 hour in ACSF at 30°C.

For slice electrophysiological recordings, the slices were transferred to a recording chamber, where they were continuously perfused with ACSF (1.5 ml/min) and bubbled with 95% O2/5% CO2 at 33°C. To record the evoked field excitatory postsynaptic potential (fEPSP) at the CA3-CA1 synapse, a glass micropipette filled with ACSF (1–2 MΩ) was placed in the stratum pyramidale of the CA1 region, and Schaffer collateral/commissure fibers were stimulated using concentric bipolar electrodes. Stimulation was performed using square pulses of 100 μs duration. The intensity of the stimulation was adjusted to produce an fEPSP amplitude that was half of the maximum. The recording signals were amplified with an Axopatch 2B amplifier, digitized with a Digidata 1320A, and recorded with Clampex 9 (Molecular Devices).

To induce long-term potentiation (LTP), we used theta-burst stimulation (TBS). TBS consists of 5 bursts of five 10-msec-duration pulses at 200-msec intervals. The baseline was determined by the average of fEPSP initial slopes from the 10-min period immediately before the TBS. The level of LTP was determined by the average of fEPSP initial slopes between from the 45 and 55 minutes after the TBS. In some experiments, 5μM bicuculline (B7561, Sigma) was applied by bath perfusion for 20 minutes before evaluating the evoked and spontaneous responses. Spontaneous responses were collected 5 minutes after the last evoked response was collected.

Locomotor activity and jumping activity

Mice were transported to the hallway outside of the testing room and habituated for one hour prior to testing. Locomotor activity was assayed using infrared activity monitors (Integra, Accuscan, Columbus, OH) in a chamber measuring 16.5 × 16.5 inches in a dimly lit room. Testing was performed between the hours of 1400 and 1700. Activity was monitored for a total of 60 minutes. For extended overnight analysis of locomotor activity, female mice aged three to six months were introduced into the same Integra instrument as described above, with access to a water bottle through the side of the chamber and food through a hole in the floor. To assess jumping activity, animals were transferred individually to a clean cage bottom. They were filmed for sixty minutes, and the number of jumps were counted manually.

Accelerating rotarod

Motor coordination was assessed using an accelerated rotarod apparatus (Rotamex-5, Columbus Instruments, OH). Mice were given a single two minute training session at constant velocity, followed by three consecutive trials per day on the accelerating rotarod for three consecutive days. Testing parameters were a starting velocity of 4 rpm, incrementing 1 rpm every ten seconds up to a maximum of 40 rpm. During testing, the trial was terminated when the mouse fell. Testing was conducted between 1300 and 1600.

Elevated plus maze

Male and female Kv2.1^+/+ and Kv2.1^−/− littermates were tested in the elevated plus maze as previously described (Workman et al., 2008). Wild-type male and female FVB mice were also tested. Each test was five minutes long and animals were scored for time spent in the open arms, closed arms, and center.

Optokinetic drum

To determine whether deletion of KV 2.1 induces gross visual deficits, groups of Kv2.1 KO mice (total= 9, f=7, m=2) and C57BL6 mice (total=9, f=8, m=1) were tested for visual head tracking using an optokinetic drum apparatus (Puk et al., 2008). Briefly, the apparatus consists of an inner acrylic cylinder (15 cm diameter × 15 cm height) that is held by a metal pole inside of a larger acrylic cylinder (56 cm diameter × 51 cm height). The inside cylinder is stationary while the outer cylinder rotates clockwise at 2.4 revolutions per minute (RPM) or 14.4 deg/sec. This speed is consistent with previous investigations in which head movements using an optokinetic drum have consistently been observed (Dooley et al., 2012, Fuller, 1985, Puk et al., 2008, Thaung et al., 2002). Lighting consisted of ambient overhead room light. The visual stimulus consisted of alternating black and white bars on the outer drum. The spatial frequency used was 0.15 cycles per degree and was calculated by determining the length of one degree of visual angle from the center of the outer cylinder, where one “cycle” consists of one black and white vertical stripe. Each mouse was placed inside the inner cylinder and allowed to acclimate to the apparatus for 2 minutes before the outer drum rotation was started. Following an additional 1 minute acclimation period with the outer drum rotating, head-tracking behavior was then recorded for each 15 second interval for 2 minutes. The observer was blind to genotype at the time of testing. A composite score for each mouse was calculated by summing all of the intervals at which head tracking behavior was observed.

Flurothyl-induced seizures

Flurothyl (bis-2,2,2-trifluoroethyl ether, Sigma-Aldrich #287571) was used for seizure induction in male Kv2.1^+/+ and Kv2.1^−/− mice (Velisek, 2006). Each animal was placed in an airtight Plexiglas chamber, and flurothyl was dripped (10 μl/min) onto filter paper from which it vaporized (Prichard et al., 1969). The latencies to the first myoclonic jerk (focal seizure), bilateral jerk, and full clonic convulsion were recorded for each mouse; these latencies constitute the seizure ‘threshold’ measurements. The observer was blind to genotype at the time of testing. At the first sign of full clonic convulsion, drug exposure was terminated by opening the chamber and removing the mouse.

Pilocarpine-induced seizures

Prior to seizure testing, mice were moved singly to clean cage bottoms and habituated to the cage and the testing room for a minimum of 60 minutes. To block peripheral side effects of pilocarpine, mice were pretreated with scopolamine methyl nitrate (Sigma #S2250) dissolved in 0.9% saline (1.0 mg/kg, i.p.). Thirty minutes later, mice were injected with the indicated dose of pilocarpine (Sigma #P6503). Pilocarpine was injected between the hours of 1200 and 1400 to minimize any circadian effects. We used a modified Racine scale (Racine, 1972) similar to that described by previously (Winawer et al., 2007) to score seizure severity: Stage 0, no abnormality; Stage 1, immobility, lying low; Stage 2, appearance of rigid posture, forelimb and/or tail extension, tremor (not continuous), head bobbing; Stage 3, continuous strong tremor and shaking; Stage 4, rearing/hyperexcitability/falling, tonic extension/clonic seizures with loss of posture; Stage 5, status epilepticus (continuous Stage 4 seizures) and/or death. The observer was blind to genotype at the time of testing. At the 200 mg/kg dose, mice up to six months of age were tested. For the lower doses, mice were between 8 and 13 weeks of age.

Electroencephalogram (EEG) recordings

Eight to sixteen week old mice were anesthetized with 4% isoflurane and placed in a mouse stereotaxic apparatus. 2–2.5% isoflurane was delivered via a mouse gas anesthesia mask, and the body temperature was held constant at 37°C via a heating pad. An incision was made to expose the skull; holes for the recording and reference electrodes were drilled at approximately 1.5 mm posterior from bregma and 1.5 mm lateral from the midline using a 0.9 mm drill bit. The wireless transmitter cap (Ripple^™, Utah, #R01057) was placed on the animal’s skull, and one lead was cemented in each hole without penetrating the dura. The transmitter cap was secured using acrylic cement that also sealed the skin around the cap. Implanted animals were placed in an empty cage on top of a wireless receiver (Ripple^™, Utah, Epoch Receiver). Signals from the receiver were funneled into Harmonie^™ DVN V5.2a EEG software for visual inspection.

Morris water maze

One week prior to the experiment, mice were housed singly to facilitate testing. All testing was performed between 0800 and 1200. The water maze pool was 86 cm (interior diameter) and filled with water made opaque with nontoxic tempura paint. Water temperature was maintained at 21°C. On the first trial of day 1, mice were placed in the pool containing a visible platform. Then an acrylic platform (6 cm × 6 cm) was submerged approximately 1.0 cm below the surface of the water in the same place. Mice were given four spaced trails per day for four consecutive days. The hidden platform remained in the same position throughout the experiment, but the starting positions were different on each trial. Inter-trial intervals were approximately 8 minutes. Each trial continued for a maximum of 90 seconds or until the mouse reached the platform. If the mouse did not reach the platform in 90 seconds, it was captured by the experimenter and gently guided to the platform where it remained for 30 seconds. Sessions were captured by an overhead camera and analyzed using the SMART v2.5 video tracking system (Harvard Apparatus, Holliston, MA). In the probe trial on Day 5, the invisible platform was removed from the pool and the mice were allowed 90 seconds to explore the water maze.

Statistics

Data are presented as mean ± SEM. Graphpad Prism ver6.0b (La Jolla, CA) was used to perform statistical analysis. Analysis of electrophysiology data was performed in Excel (Microsoft, USA).

Results

Kv2.1^−/− mice are grossly normal and do not show evidence of compensatory Kv channel expression

To assess the impact of Kv2.1 deletion on the mouse brain we looked for evidence of compensatory changes in the anatomy, biochemistry, and electrophysiology of Kv2.1 mutant mice (official allele designation Kcnb1^tm1Dgen, hereafter referred to as Kv2.1^−/−). Multiple label immunofluorescence of hippocampal brain sections from wild-type (Kv2.1^+/+) and Kv2.1^−/− mice revealed no changes in gross hippocampal neuroanatomy. In Kv2.1^+/+ brain tissue, Kv2.1 immunolabeling is present throughout the hippocampus, with a high level of expression in CA1 pyramidal neurons, in dentate granule cells, and in subicular pyramidal neurons, as well as in interneurons scattered throughout the various hippocampal subfields; CA2 and CA3 pyramidal neurons showed lower levels of expression (Fig. 1a). Each of these features of Kv2.1 immunolabeling is eliminated in sections prepared from Kv2.1^−/− littermates (Fig. 1b). In contrast, immunolabeling for other Kv channels, and overall hippocampal cytoarchitecture as assessed by labeling nuclei with a DNA stain, remain grossly unaltered in Kv2.1^−/− mice. As one example, the characteristic immunolabeling of the Kv1.4 α subunit that is observed in brain sections from rats (Cooper et al., 1998, Monaghan et al., 2001) and Kv2.1^+/+ mice (Misonou et al., 2006a) is also retained in sections from Kv2.1^−/− mice (Fig. 1a, b). Immunolabeling for the highly related Kv2.2 α subunit is also grossly unaffected in sections from Kv2.1^−/− mice (Figs. 1c and 1d).

Figure 1. Lack of apparent compensatory changes in Kv channel expression in Kv2.1^−/− mice.

Multiple immunofluorescence labeling of adult hippocampus for (a–d). Panels (a, c) are from Kv2.1^+/+ mice, and panels (b, d) from Kv2.1^−/− littermates. Sections were labeled for Kv2.1 (green) and (a–b) Kv1.4 (red), or (c–d) Kv2.2 (red), and directly labeled with a nuclear dye (blue), as indicated. Note that Kv2.1 staining is absent in Kv2.1^−/− brain sections and that the expression and localization of Kv1.4 and Kv2.2 seen in the Kv2.1^+/+ section is maintained in the Kv2.1^−/− section. Scale bar: 200 μm. (e) Representative immunoblots of Kv channel α subunit expression in whole brain of Kv2.1^+/+ and Kv2.1^−/− adult mice. (f) Quantitation of protein expression levels from nine Kv2.1^+/+ (filled bars) and ten Kv2.1^−/− (open bars) MBM fractions. The graph depicts the average intensity normalized to Kv2.1^+/+ for each antibody used. Data are expressed as mean ± SEM. ****P < 0.00001.

Immunoblots from whole adult brain crude membrane preparations reveal a similar loss of Kv2.1 expression, and a lack of significant compensatory changes of other Kv channel α subunits in the brains of Kv2.1^−/− mice (Fig. 1e, f). Representative data from immunoblots performed with an anti-Kv2.1 monoclonal antibody reveal robust immunoreactivity in crude membrane samples prepared from the brains of Kv2.1^+/+ mice, but no detectable anti-Kv2.1 immunoreactivity in similar samples prepared from Kv2.1^−/− littermates. Fig. 1e also shows representative data from immunoblots on these samples with a panel of monoclonal antibodies specific for other brain Kv channel α subunits, and for the Hsp70 family member mortalin/GRP75, which is highly expressed in brain (Kaul et al., 1997). Quantification of immunoblots from a total of nine Kv2.1^+/+ and ten Kv2.1^−/− mice are shown in Fig. 1f and reveals an absence of Kv2.1 protein in Kv2.1^−/− mice (t₁₇=10.7, P<0.00001) but no significant differences in the total protein expression of Kv1.2, Kv1.4, Kv1.5 and Kv2.2 α subunits between Kv2.1^+/+ and Kv2.1^−/− mice. Together, these results confirm the high level of expression of Kv2.1 immunoreactivity in neurons of Kv2.1^+/+ mice, and the elimination of this expression in the brains of their Kv2.1^−/− littermates. Notably, Kv2.1 deletion did not induce significant changes in the gross anatomy of the hippocampus, or in the expression and localization of other Kv channels α subunits in this region.

Slowly deactivating delayed rectifier current is reduced in Kv2.1^−/− neurons

To determine whether Kv2.1 deletion leads to loss of neuronal K⁺ current, whole cell patch clamp recordings were performed on hippocampal pyramidal neurons cultured from Kv2.1^+/+ and Kv2.1^−/− littermates. In doing so we utilized an experimental preparation where the loss of Kv2.1 was unlikely to be compensated for by changes in other K⁺ currents, as studies of other Kv subunit knockouts have been confounded by electrical remodeling (Guo et al., 2005, Nerbonne et al., 2008). To minimize potential compensation for loss of this channel subunit, K⁺ currents were recorded from neonatal (P0) cultures at 4–6 days in vitro (DIV), as this is the earliest stage in culture during which cell surface expression of Kv2.1 was observed in similar cultures from rats (Antonucci et al., 2001). Neurons from both Kv2.1^+/+ and Kv2.1^−/− mice yielded substantial I_K (Fig. 2a, b; black traces), indicating that channels other than Kv2.1 dominate I_K at this early developmental stage. Nonetheless, electrophysiological effects of Kv2.1 deletion were apparent. To assess the contribution of Kv2.1 to neuronal currents pharmacologically, we used Guangxitoxin (GxTX), a potent inhibitor of Kv2 channels (Herrington et al., 2006). As expected, GxTX inhibited currents from Kv2.1^+/+ neurons (Fig. 2a, c; gray). In addition to Kv2.1, GxTX also inhibits Kv2.2 and modulates Kv4 family channels (Herrington et al., 2006), and these channels could account for the GxTX sensitive currents remaining in some Kv2.1^−/− neurons (Fig. 2b, c). To better distinguish the contribution of Kv2.1, we isolated the notably slow deactivation of Kv2.1 channels (Shieh et al., 1997, Taglialatela & Stefani, 1993) with a tail current protocol (Fig. 2a). The slowly deactivating component of GxTX-sensitive tail currents (Fig. 2d) had voltage dependent characteristics similar to heterologously expressed Kv2.1 (Shieh et al., 1997, Taglialatela & Stefani, 1993). This prominent slow tail current was sufficient to distinguish Kv2.1^+/+ neurons from Kv2.1^−/− (Mann-Whitney U=0.0, P<0.001; Fig. 2f). These data reveal that a component of neuronal I_K with the expected properties of Kv2.1 was reduced by Kv2.1^−/− deletion.

Figure 2. A slowly deactivating delayed rectifier current is reduced in Kv2.1^−/− neurons.

Representative current traces from (a) Kv2.1^+/+ and (b) Kv2.1^−/− mouse hippocampal neurons. Cell capacitance was 22pF and 20pF respectively. Voltage clamp command indicated above. Black traces, currents in BSA vehicle. Gray traces, currents in presence of 100 nM GxTX. Blue and red traces, subtraction of GxTX from vehicle trace. Box indicates region of slow tail currents used for analysis. (c) Peak currents normalized to cell capacitance from Kv2.1^+/+ and Kv2.1^−/− neurons during step to 10 mV. Black circles, vehicle. Gray circles, 100nM GxTX. Lines connect data from individual neurons. Thick bar is mean, error bars SEM. (d) Magnification of subtraction currents normalized to cell capacitance from panels a and b. Smooth lines are single exponential decays fit to currents. τ^+/+ = 57.9 ± 0.2 ms, τ^−/− = 31.7 ± 0.5, mean ± SD. (e) Stimulus dependence of slow tail currents from Kv2.1^+/+, blue, and Kv2.1^−/− neurons, red. Values plotted are mean current normalized to cell capacitance from 20 to 100 ms after start of step to −40 mV from indicated stimulus voltage. Smooth lines are Boltzmann distributions fit to currents. V_1/2^+/+ = −26 ± 2 mV, z^+/+ = 4 ± 1 e, V_1/2^−/−, = −17 ± 7 mV, z^+/+ = 3 ± 2 e, mean ± SD. (f) Circles are mean of GxTX-sensitive slow tail currents from individual neurons following step from 10 mV. Values were normalized to vehicle tail following 80 mV stimulus. Kv2.1^+/+ n = 6, Kv2.1^−/− n = 9. ***P < 0.001.

At this early developmental time point, we did not see compensation for the loss of Kv2.1. The I_K in the presence of the Kv2.1 inhibitor did not show any obvious differences between Kv2.1^+/+ and Kv2.1^−/− neurons (Fig 2a, b, c). To further inspect for potential electrical remodeling due to Kv2.1 deletion, Kv currents were examined using a more negative holding potential to stimulate fast-inactivating A-type K⁺ currents in addition to delayed rectifier currents. With all stimulus protocols tested, any differences in peak, sustained, or tail currents between wild type and Kv2.1^−/− neurons were eliminated by GxTX (Fig S2). In the presence of the Kv2.1 inhibitor, the lack of difference between K⁺ currents from Kv2.1^+/+ and Kv2.1^−/− neurons suggests a lack of Kv current remodeling upon Kv2.1 deletion.

Kv2.1^−/− neurons are hyperexcitable and long-term potentiation is diminished

Given the prominent expression of Kv2.1 in hippocampus (Fig. 1), and the loss of a component of I_K in cultured hippocampal neurons from Kv2.1^−/− mice (Fig. 2), we next looked at the effects of ablation of Kv2.1 on basic properties of hippocampal neuronal excitability and circuitry. Given the high level expression of Kv2.1 in CA1 pyramidal neurons, we focused on this hippocampal subfield. Extracellular field recordings were made from the CA1 region of hippocampal slices from Kv2.1^+/+ and Kv2.1^−/− mouse brain slices submerged in artificial cerebrospinal fluid (ACSF). Representative traces of evoked responses to Schaffer collateral stimulation in Kv2.1^+/+ (left) and Kv2.1^−/− (right) slices are shown in Fig. 3a. While the duration and amplitude of evoked responses were not significantly different between Kv2.1^+/+ and Kv2.1^−/− hippocampal slices in control ACSF (Fig. S3a), differences in neuronal excitability were observed when the tissue was treated with bicuculline, a GABA_A receptor antagonist that reduces inhibition in this circuit. Responses in Kv2.1^−/− hippocampal slices were increased in the presence of 5μM bicuculline (duration: t₁₄=2.19, P<0.05; amplitude: t₁₄=2.31, P<0.05; Fig. 3b). Kv2.1^−/− slices exhibited a large, late, long-lasting event not seen in the Kv2.1^+/+ tissue—suggestive of prolonged discharge.

Figure 3. Kv2.1^−/− hippocampal brain slices have altered excitability and neuronal network function.

(a and b) Bicuculline-induced excitability is enhanced in hippocampal slices from Kv2.1^−/− mice in ACSF (5 mM KCl). (a) Representative evoked field potential traces from the CA1 region of hippocampal slices prepared from Kv2.1^+/+ and Kv2.1^−/− littermate mice before (control) and after bicuculline (5 μm bicuculline) application. Bicuculline application led to bursting in response to Schaffer collateral orthodromic stimulation in both genotypes; however, the response was more prolonged in the slices from Kv2.1^−/− compared to littermate Kv2.1^+/+ mice. (b) Summary of the increase in duration and amplitude of the evoked response after bicuculline application (as a percent of control) for slices from both Kv2.1^+/+ (n = 9 slices from 3 mice) and Kv2.1^−/− (n = 8 slices from 3 mice). (c, d and e) Excitability differences during the theta burst stimulation (TBS). (c) A representative trace of evoked epileptiform activity (E.A.) observed in Kv2.1^−/− slices (this trace 60 min after TBS). (d) Representative traces during TBS from Kv2.1^+/+ and Kv2.1^−/−, and Kv2.1^−/− slices with spontaneous epileptiform activity (EA). Each trace superimposes 5 bursts (5 pulses, 100 Hz). (e) Summary bar graph of the negative field potential area during TBS (calculated as the field potential area beneath the dotted lines in Fig. 3d). The area of the Kv2.1^+/+ (23.5±1.6 mV · ms, n=9 slices from 7 mice) was significantly larger than the area from of the Kv2.1^−/− (18.5±1.5 mV · ms, n=8 slices from 5 mice). The area of the Kv2.1^−/− slices with epileptiform activity (EA) (44.1±4.4 mV · ms, n=8 slices from 4 mice) is significantly larger than the area of Kv2.1^+/+ and Kv2.1^−/− slices. (f and g) LTP induced by TBS in Kv2.1^+/+ and Kv2.1^−/− slices in ACSF (5mM KCl). Upper traces of (f) show representative traces before TBS (control, gray line) and 55 min. after TBS (black line) from Kv2.1^+/+ and Kv2.1^−/− hippocampal slices. Bottom of (f) indicates the time course of normalized fEPSP slope as mean ± SEM (Kv2.1^+/+, n = 9 slices; Kv2.1^−/−, n = 8 slices). LTP at the CA3-CA1 synapse was induced by TBS at time point 0 (indicated by black arrow), and monitored for 60 min. after induction. TBS induced LTP in both Kv2.1^+/+ and Kv2.1^−/− hippocampal slices. However, LTP amplitude was smaller in Kv2.1^−/− slices at 45–55 min after TBS. (g) Summary bar graph indicates that TBS induced LTP in Kv2.1^+/+ and Kv2.1^−/− hippocampal slices. The magnitude of LTP was significantly different between Kv2.1^+/+ (195%±10%, n=9 slices from 7 mice) and Kv2.1^−/− (151%±4%, n=8 slices from 5 mice). Data are expressed as mean ± SEM. *P < 0.05; ****P<0.0001.

We next tested the effects of loss of Kv2.1 on the induction of long-term potentiation (LTP) in the CA3-CA1 Schaffer collateral pathway in hippocampal slices. The Schaffer collateral pathway was stimulated and field EPSPs (fEPSPs) were recorded in stratum radiatum of CA1. LTP measurements were complicated by epileptiform activity (EA) in Kv2.1^−/− slices (e.g. see Fig. 3a). In response to test pulses, approximately 40% of slices from Kv2.1^−/− mice exhibited EA not seen in slices prepared from Kv2.1^+/+ mice. A representative trace of this epileptiform activity is shown in Fig. 3c. During theta burst stimulation (TBS), the negative field potential area of Kv2.1^−/− slices exhibiting EA (Kv2.1^−/− EA) was higher than that of Kv2.1^+/+ slices (field potential beneath the dotted lines in Fig. 3d). The penetrance of the EA phenotype was incomplete, as there existed a second group of Kv2.1^−/− slices that did not exhibit EA and had a lower negative field potential area during TBS than Kv2.1^+/+ slices (Kv2.1^+/+ vs. Kv2.1^−/−: t₈₃=2.26, P<0.05; Kv2.1^+/+ vs. Kv2.1^−/− EA: t₈₃=9.25, P<0.0001; Kv2.1^−/− vs. Kv2.1^−/− EA: t₇₈=17.17, P<0.0001; Fig. 3e). The basis of the distinction in epileptiform activity between these two populations of Kv2.1^−/− hippocampal slices is not clear, as it is independent of animal from which the slices were prepared, the litter from which the mice were derived, the time between when the brain was excised and the slices prepared, and the time of the experiment. For LTP experiments, Kv2.1^−/− EA slices were excluded from the analysis because control experiments revealed that EA in wild type slices also interferes with the development of LTP (Fig. S3b). Although TBS induced LTP in both Kv2.1^+/+ and Kv2.1^−/− slices, LTP amplitude was significantly lower in Kv2.1^−/− slices (Fig. 3f), as quantitated in Fig. 3g (Kv2.1^+/+: t₁₈=20.13, P<0.0001; Kv2.1^−/−: t₁₈=14.89, P<0.0001; Kv2.1^+/+ vs. Kv2.1^−/−: t₁₅=1.64, P<0.05). Experiments using 3 mM KCl instead of 5 mM KCl to reduce overall excitability yielded qualitatively similar results (Fig. S3c). In summary, the loss of Kv2.1 expression alters neuronal excitability and plasticity in the hippocampus. This altered hippocampal network function suggests that behaviors linked to hippocampal function may be perturbed

Lower weight but normal motor coordination and visual acuity in Kv2.1^−/− mice

While grossly normal in appearance and behavior, Kv2.1^−/− male mice weigh significantly less than age-matched Kv2.1^+/+ littermates at all ages tested, a difference that became more pronounced after approximately three months of age (two-way RMANOVA, F_1,108(genotype)=15.51, P<0.01; Fig. 4a). The weight of Kv2.1^+/− (heterozygous) male mice was not significantly different from age-matched Kv2.1^+/+ males (data not shown). There was not a significant difference in weight between Kv2.1^−/− and Kv2.1^+/+ female mice until about three months of age, when the weight of Kv2.1^+/+ females became significantly higher than that of the mutants (two-way RMANOVA, F_1,144(genotype)=1.85, P>0.05). To minimize complications of weight differences in subsequent behavioral assays, most behavioral tests were performed between the ages of two and three months, prior to when substantial weight differences were observed. We first assessed motor coordination using an accelerating rotarod assay. Kv2.1 is expressed the cerebellum (Drewe et al., 1992), as well as in spinal motor neurons (Muennich & Fyffe, 2004) and skeletal muscle (J.S.T., personal observation), suggesting that motor defects could result from constitutive and global deletion of Kv2.1. We found no differences in performance between Kv2.1^+/+ and Kv2.1^−/− males (two-way RMANOVA, F_1,144(genotype)=0.24, P>0.05; Fig. 4b). As Kv2.1 is expressed in the retina (Klumpp et al., 1995), potential impacts on visual acuity were assessed. We observed no differences between Kv2.1^+/+ and Kv2.1^−/− littermates in vision as determined by head tracking in an optokinetic drum (Fig. 4c), used to examine visual properties in mice (Puk et al., 2008).

Figure 4. Basic assessment of Kv2.1^−/− mice.

(a) Kv2.1^+/+ and Kv2.1^−/− male mice were weighed weekly. Kv2.1^+/+ male mice weighed significantly more than Kv2.1^−/− at all time points, particularly after three months of age (n=7/genotype). Note that the SEM (error bars) is included in these graphs; there was very little variation in weight within a genotype. (b) Rotarod performance is unaffected in Kv2.1^+/+ and Kv2.1^−/− male mice (n=10/genotype). (c) Optokinetic head tracking response is unaffected Kv2.1^−/− female mice (n=7) relative to Kv2.1^+/+ littermates (n=8). Data are expressed as mean ± SEM. *P < 0.01.

Kv2.1^−/− mutant mice are strikingly hyperactive

General locomotor activity was monitored over a twenty-four hour period, revealing greatly increased activity in Kv2.1^−/− mice, with particularly high spikes of activity following transitions in lighting (Fig. 5a). We also found that Kv2.1^−/− mice were hyperactive when challenged with a novel environment. Locomotor activity of Kv2.1^−/− mice over a sixty-minute period was nearly three-fold higher than that of Kv2.1^+/+ mice (two-way RMANOVA, F_1,342(genotype)=40.26, P<0.0001; Fig. 5b) and (t₁₈=6.34, P<0.0001; Fig. 5c). The hyperactivity of the knockouts also manifests in a stereotypical jumping behavior after a cage change. Kv2.1^−/− mice jumped repetitively in their cages, with nearly 95% of females and 60% and males exhibiting this behavior (Mann-Whitney U=3.0, P<0.0001; Fig 5d), which can occur in bouts of up to 4,300 jumps/hour.

Figure 5. Kv2.1^−/− mice are hyperactive.

(a) 24 hour locomotor activity was elevated in Kv2.1^−/− female mice (n=7) relative to Kv2.1^+/+ littermates (n=8). Each data point represents seven minutes of activity (cm/7 minutes). (b) Spontaneous locomotor activity is significantly higher Kv2.1^−/− male mice than in Kv2.1^+/+ littermates (n=10/genotype). Each data point represents three minutes of activity (cm/3 minutes). (c) Cumulative Distance Traveled (Cumulative DT) over the sixty-minute period in B showed a nearly Mendelian separation between the genotypes. (d) Female mice were transferred to a new cage bottom and the number of jumps on the wall of the cage was counted over a one hour time period. Data are expressed as mean ± SEM. ***P < 0.0001.

Reduced anxiety-like behavior in Kv2.1^−/− mutant mice

Kv2.1^−/− mice also exhibited significantly less anxiety-like behavior in an elevated plus maze. This phenotype was observed in both genders and in two separate testing facilities with different experimenters. Kv2.1^−/− mice spent a significantly higher percentage of their time exploring the open arms of the plus maze than Kv2.1^+/+ littermates (males: F_2,54(genotype × location)=18.2, P<00001, Fig. 6a; females: F_2,42(genotype × location)=47.2, P<0.0001, Fig 6c). Furthermore, Kv2.1^−/− mice entered the open arms of the plus maze significantly more often than Kv2.1^+/+ littermates (males: Mann-Whitney U=13.5, P=0.004, Fig. 6b; females: Mann-Whitney U=6.5, P=0.022, Fig. 6d).

Figure 6. Decreased anxiety-like behavior in Kv2.1^−/− mice.

Percent time spent in open arms of elevated plus maze (a and c) and number of entries into the open arms of the maze (b and d) are significantly increased in Kv2.1^−/− mice compared to Kv2.1^+/+ littermates for both male mice (a and b; n=10/genotype) and female mice (c and d; n=8 Kv2.1^+/+ and n=6 Kv2.1^−/− mice). Data are expressed as mean ± SEM. *P < 0.05; **P < 0.01; ***P < 0.0001.

Kv2.1^−/− mutant mice lack spatial learning ability

Given the reduced level of hippocampal LTP in the Kv2.1^−/− mice (Fig. 3), we next assayed whether these mice had deficits in learning and memory. Spatial learning was assessed in a Morris Water Maze test, where after exposure to a visible platform, Kv2.1^−/− and Kv2.1^+/+ mice were challenged to find a hidden platform over a period of four training days. We observed a significant difference in the acquisition of this task between the two genotypes as a function of time (two-way RMANOVA, F_3,42(genotype × time)=3.21, P<0.05; Fig. 7a), with Kv2.1^−/− mice performing significantly worse than Kv2.1^+/+ littermates. Based on the latencies to find the platform between Day 1 and Day 4, Kv2.1^−/− mice did not demonstrate any evidence of spatial learning (t₇=0.06, P=0.95; Fig. 7a). In contrast to the marked locomotor hyperactivity noted above, swim speed was not significantly different between the mice of differing genotype (t₁₄=0.19, P=0.85; Fig. 7b). When the platform was removed during the probe trial, the average distance from the platform location was significantly lower (t₁₄=2.27, P<0.05; Fig. 7c) and the number of platform crossings was higher (t₁₄=4.07, P<0.01; Fig 7d) in Kv2.1^+/+ relative to Kv2.1^−/− mice. Although Kv2.1^−/− mice are somewhat prone to seizures (see below), we observed only two incidents of brief seizure-like activity in the Morris Water Maze experiment: one in a Kv2.1^+/+ mouse and the other in a Kv2.1^−/− mouse (both on the very first training trial). Thus, seizures per se did not appear to account for poor learning in Kv2.1^−/− mice.

Figure 7. Kv2.1^−/− mice are defective in learning in a Morris Water Maze.

(a) Latency to find the hidden platform did not decrease significantly over the testing days in Kv2.1^−/− male mice but did in Kv2.1^+/+ littermates. During the probe trial on Day 5, swim velocity did not differ between Kv2.1^+/+ and Kv2.1^−/− male mice (b) However, the distance from the target platform was significantly higher for Kv2.1^−/− mice (c), and the number of platform crossings was significantly lower (d) (n=8/genotype for all experiments). Data are expressed as mean ± SEM. *P < 0.05; **P < 0.01.

Seizure suppression is impaired Kv2.1^−/− mice

The activity of Kv2.1 as a homeostatic suppressor of excitotoxic signaling in central neurons suggests Kv2.1^−/− mice might be prone to seizures. During routine animal handling (cage changes, weighing, etc.), we noted that approximately 10% of Kv2.1^−/− mice experience non-lethal, tonic-clonic seizures lasting several seconds (D.J.S., personal observation). To investigate this further, seizure susceptibility of Kv2.1^−/− mice was assessed using two chemical convulsants, flurothyl and pilocarpine. No significant differences were observed between Kv2.1^+/+ and Kv2.1^−/− mice exposed to the inhalant convulsive flurothyl in the latency to the first myoclonic jerk, the initial stage of the seizure response (Fig. 8a). However, the transitions from the first myoclonic jerk to the more severe seizure stages of bilateral clonus and tonic-clonic seizures were significantly faster in the Kv2.1^−/− mice than in Kv2.1^+/+ mice (myoclonic to bilateral: t₁₂=3.11, P<0.01; myoclonic to tonic-clonic: t₁₂=2.76, P<0.05; Fig. 8b), occurring over a period of less than 30 seconds, whereas Kv2.1^+/+ mice required minutes to progress through these stages (Fig. 8b). Thus, while elimination of Kv2.1 did not impact the initial behavioral response to flurothyl exposure, it did enhance the progression to the more severe seizure stages.

Figure 8. Kv2.1^−/− mice are susceptible to chemically induced seizures.

(a) Latency to flurothyl-induced seizures was not significantly different between Kv2.1^+/+ and Kv2.1^−/− male mice, except for latency to tonic-clonic seizure (n=7/genotype). (b) Transition duration from one seizure stage to another was significantly faster in Kv2.1^−/− male mice relative to Kv2.1^+/+ littermates (n=7/genotype). (c) Incidence to pilocarpine-induced seizures (Stage 4, see Methods) was significantly increased in Kv2.1^−/− and Kv2.1^+/− male mice relative to Kv2.1^+/+ littermates at all doses (n=8/dose/genotype). (d) Latency to pilocarpine-induced seizures (Stage 4) was significantly decreased in Kv2.1^−/− and Kv2.1^+/− male mice relative to Kv2.1^+/+ littermates at all doses (n=8/dose/genotype). (e) Representative EEG recordings from Kv2.1^+/+ and Kv2.1^−/− male mice confirmed that pilocarpine (170mg/kg) induced ictal patterns of activity in Kv2.1^−/− animals. Data are expressed as mean ± SEM. *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.001.

We also determined latency to and incidence of seizures using different doses of the muscarinic receptor agonist pilocarpine. We chose a starting dose of pilocarpine that was unlikely to induce seizures in Kv2.1^+/+ C57BL/6J animals, based on previous studies (Schauwecker, 2012, Winawer et al., 2011, Winawer et al., 2007). Seizure stages were scored according to the criteria outlined in Table 1. Because lethality due to pilocarpine injection is very high on the C57BL/6J genetic background (Schauwecker, 2012), we monitored the latency to and incidence of severe Stage 4 seizures (rearing/hyperexcitability/falling, tonic extension/clonic seizures with loss of posture) rather than status epilepticus. We found that Kv2.1^−/− mice had a higher incidence (Fisher’s exact test, 150mg/kg: P<0.001; 200mg/kg: P<0.05; Fig. 8c) and a shorter latency (200mg/kg: t₉=6.94, P<0.0001; Fig. 8d) to Stage 4 seizures than Kv2.1^+/+ littermates, and that this effect was dose-dependent. We also determined that heterozygous Kv2.1^+/− mice had seizure phenotype intermediate between that of Kv2.1^+/+ and Kv2.1^−/− mice (150mg/kg, Kv2.1^−/− vs. Kv2.1^+/−: t₈=4.86, P<0.01; Fig. 8d) suggesting that the level of Kv2.1 expression influences seizure susceptibility. We also found that in response to pilocarpine injections Kv2.1^−/− mice had significantly higher incidences of both status epilepticus (SE) and mortality relative to Kv2.1^+/+ mice (Table S1). Although we did not detect epileptiform activity in electroencephalographic (EEG) monitoring of Kv2.1^−/− mice during a 24 hour free running period (E.T.D. and P.A.S., personal observations), we confirmed that pilocarpine was, in fact, inducing epileptiform EEG activity (and not simply abnormal motor behavior) in Kv2.1^−/− mice at doses that did not elicit any epileptiform EEG activity in Kv2.1^+/+ littermates (170 mg/kg pilocarpine in Fig. 8e). These results confirm that Kv2.1^−/− mice are more susceptible to chemically induced seizures than are Kv2.1^+/+ mice. Kv2.1 deletion accelerates seizure progression, presumably as a neurological consequence of the loss of a homeostatic suppressor of neuronal electrical activity.

Table 1.

Limbic seizure stages in pilocarpine-treated mice^a

Stage 0	No abnormality
Stage 1	Immobility. Lying low.
Stage 2	Rigid posture. Forelimb and/or tail extension. Tremor (not continuous). Head bobbing.
Stage 3	Continuous strong tremor/shaking.
Stage 4	b Rearing/hyperexcitability/falling, tonic extension/clonic seizures with loss of posture.
Stage 5	Status epilepticus (continuous Stage 4 seizures) and/or death.

^aSeizure criteria derived from Winawer, et. al., 2007.

^bTonic seizures are characterized by whole-body stiffening and extension. Clonic seizures are characterized by repetitive rhythmic jerking. These two phenomena are typically combined in whole body convulsive episodes.

Discussion

Kv2.1 is a critical regulator of intrinsic excitability and a key mediator of I_K in the nervous system. Given its high levels of expression, elimination of Kv2.1 might be expected to have widespread impact on behavior. Indeed, we have observed a number of striking phenotypes in Kv2.1^−/− mice, including defects in spatial learning, locomotor hyperactivity, reduced anxiety-like behavior, and increased sensitivity to chemoconvulsant-induced seizures.

We found immunohistochemical and electrophysiological evidence of Kv2.1 loss and saw no evidence of compensation. We confirmed the loss of Kv2.1 expression in the brains of Kv2.1^−/− mice and determined that there was no apparent change in the protein density of several other Kv channels (Fig. 1), although we cannot rule out the possibility of other compensatory changes in the mutant mice, which could include developmental compensation or changes in expression or surface expression of other gene products, alterations in subunit stoichiometry, phosphorylation or other secondary modifications, localization within the neurons, alterations in synaptic connectivity and/or efficacy, etc. The lack of obvious compensatory changes in other Kv channel α subunits is grossly similar to what has been observed in other Kv channel KO mice [as reviewed in (Menegola et al., 2012)]: for example mice lacking expression of Kv1.1 (Smart et al., 1998, Wenzel et al., 2007), and Kv4.2 (Chen et al., 2006, Menegola & Trimmer, 2006). However, changes in expression of auxiliary subunits have been observed in KO mice, such as those normally affiliated with Kv4.2 α subunits (Chen et al., 2006, Menegola & Trimmer, 2006). A number of previous studies have demonstrated that Kv2.1 channels are responsible for the bulk of I_K in hippocampal neurons. Antisense oligonucleotide-mediated knockdown of Kv2.1 yielded a >50% reduction in sustained outward K⁺ current in hippocampal slices prepared from P7 rats and cultured for up to 19 days in vitro or DIV (Du et al., 2000). Antibody block of Kv2.1 yielded a similar >50% reduction in whole cell sustained outward K⁺ current in dissociated hippocampal neuronal cultures prepared from E19 embryos and cultured for 14 DIV (Murakoshi & Trimmer, 1999). We chose to measure Kv2.1 currents soon after their first expression, to minimize interference by pharmacologically indistinguishable Kv channel α subunits, such as Kv2.2. As expected from the early developmental stage of mouse neurons used in this study, the loss of I_K in Kv2.1^−/− neurons was less obvious than seen in these previous studies with more advanced rat neurons. In E19 rat neuron cultures, intracellular/endoplasmic reticulum associated immunolabeling was seen at 4 DIV, and plasma membrane expression of Kv2.1 was seen at 6 DIV (Antonucci et al., 2001), suggesting this to be the earliest time window in which current from Kv2.1 might be seen. We measured currents from P0 mouse hippocampal neuron cultures at 4–6 DIV, the expected onset of Kv2.1 expression. At this early time point a significant decrease in a current with Kv2.1-like properties was observed in Kv2.1^−/− mice (Fig. 2), without confounding electrical remodeling of Kv currents (Fig. S2). As Kv2.1 expression increases at later developmental stages, more Kv2.1 derived I_K is expected in Kv2.1^+/+ along with possibility of electrical compensation in Kv2.1^−/− animals.

The difference in body weight between Kv2.1^−/− and Kv2.1^+/+ littermates (Fig. 4a) could be a consequence of behavioral or hormonal changes. Previous studies using this same Kv2.1^−/− line have demonstrated enhanced glucose tolerance and insulin secretion in Kv2.1^−/− mice (Jacobson et al., 2007), which could contribute to their reduced weight. However, other alterations in eating behavior or metabolism could also contribute to the phenotype. For instance, Kv2.1^−/− mice are strikingly hyperactive relative to Kv2.1^+/+ littermates, exhibiting a near Mendelian segregation based solely on this phenotype. This hyperactivity is clearly observed when the animals are placed in a novel environment (Fig. 5b, c), although it is also possible that Kv2.1^−/− mice are also hyperactive in their home cage, which might explain the weight difference. Kv2.1^−/− mice also exhibit repetitive jumping and rearing when transferred to a new cage bottom (Fig. 5d). The striking hyperactivity in a novel environment and perseverative jumping behavior of Kv2.1^−/− animals is reminiscent of that seen in mice lacking the dopamine transporter (Giros et al., 1996, Ralph et al., 2001), and consistent with a role for Kv2.1 in homeostatic suppression of elevated neuronal activity. Interestingly, Kv2.1 was identified as a dopamine transport interacting protein, suggesting a potential role in dopaminergic neurons (Maiya et al., 2007). Kv2.1 is also responsible for a slowly deactivating Kv current in globus pallidus neurons, where it has been proposed to impact their discharge rate (Baranauskas et al., 1999), and medium spiny neurons in mouse models of Huntington disease have reduced levels of Kv2.1 expression (Ariano et al., 2005), suggesting an important role for this delayed rectifier Kv channel as a conditional regulator of neuronal activity in basal ganglia circuitry.

In addition to being hyperactive, another striking phenotype exhibited by Kv2.1^−/− animals is a significant increase in time spent in the open arms in the elevated plus maze (Fig. 6). This phenotype was observed in both male and female animals and was observed independent of facility or experimenter. It is important to note that although Kv2.1 is expressed in the retina (Klumpp et al., 1995), optokinetic drum experiments uncovered no deficit in head tracking behavior in Kv2.1^−/− mice, suggesting that the differences in performance in the elevated plus maze (and the Morris Water Maze) presumably do not arise from differences in visual acuity.

Kv2.1^−/− animals do not exhibit spontaneous seizures, yet induced seizures progress more rapidly. This contrasts with Kv1.1^−/− (Smart et al., 1998, Wenzel et al., 2007) and Kv1.2^−/− (Brew et al., 2007) mice, which exhibit spontaneous seizures that can be fatal (Robbins & Tempel, 2012). Kv2.1^−/− mice do experience handling-induced seizures, but at a low frequency (~10%), and these seizures only last for several seconds and are never lethal. On the other hand, we found that Kv2.1^−/− mice are susceptible to chemically induced seizures (Fig. 8). While the elimination of Kv2.1 did not impact the latency to initial seizure activity for flurothyl, the transition time between myoclonic jerk and bilateral clonus (and between myoclonic jerk and tonic-clonic seizure) was significantly faster in the Kv2.1^−/− mice; that is, seizure generalization proceeded more rapidly. Kv2.1^−/− mice also exhibit behavioral and electrographic seizures in response to lower doses of pilocarpine that do WT mice. These observations may be consistent with the role of Kv2.1 as a conditional regulator of neuronal excitability (Du et al., 2000, Misonou et al., 2005, Mohapatra et al., 2009, Surmeier & Foehring, 2004). Mice lacking Kv2.1 are presumably missing this important homeostatic suppression mechanism, which in Kv2.1^+/+ mice is mediated via activity-dependent and seizure-induced Kv2.1 dephosphorylation. The resultant recruitment of Kv2.1 channel activity yields homeostatic suppression of neuronal excitability, which could act to dampen the spread of the seizure and its severity, such that Kv2.1^−/− mice lacking this mechanism display this heightened response to chemically induced seizures. Deletion of Kv2.1 expression in specific neuronal populations will provide insights into the underlying mechanisms of the intriguing enhanced progression of the Kv2.1^−/− mice through the later seizure stages, and the sensitivity to low dose pilocarpine seizures.

The fact that Kv2.1^−/− animals are defective in learning the Morris Water Maze task (Fig. 7) may be related to the LTP defects seen in Kv2.1^−/− brain slices (Fig. 3). The mechanism whereby LTP is deficient in slices from Kv2.1^−/− mice is not immediately apparent. In Kv2.1^−/− slices, approximately 40% of the slices were hyperexcitable in response to a test pulse (Fig. 3), a result that was not unexpected given the role of Kv2.1 as a conditional suppressor of neuronal excitability. In the other 60% of the Kv2.1^−/− hippocampal slices that were not hyperexcitable, however, LTP was also reduced compared to Kv2.1^+/+ littermates. There are several possible explanations for reduced LTP in the Kv2.1^−/− mice. First, it is possible that epileptiform activity (EA) is present in all Kv2.1^−/− brain slices but was not detected during our observations of some slices. Occurrence of EA prior to the recordings (e.g., in vivo) could disrupt the later development of experimentally induced LTP. Second, while antisense oligonucleotide-mediated knockdown Kv2.1 yielded enhanced dendritic calcium transients (Du et al., 2000), which generally support enhanced LTP, such alterations could lead to disruption of LTP through a distinct, and perhaps more complex, mechanism. To underscore this, deletion of a different dendritic Kv channel α subunit, Kv4.2, results in defects in learning in the Morris Water Maze (Lockridge & Yuan, 2011, Lugo et al., 2012), yet Kv4.2 KO mice exhibit increased LTP, presumably due to increased backpropagating action potentials (Chen et al., 2006). Future experiments will address the mechanism whereby elimination of Kv2.1 impacts synaptic plasticity associated with LTP. Last, Kv2.1 is expressed in both principal cells and inhibitory interneurons (Fig. 1) (Du et al., 1998). Stimulation of Schaffer collaterals would excite both principal neurons of the CA1 region as well as surrounding inhibitory interneurons. Loss of Kv2.1 in these interneurons could lead to greater activity, enhanced GABA release and inhibition of the LTP. The widespread expression of Kv2.1 in principal cells and interneurons throughout the hippocampus (Fig. 1) suggests that could have complex effects on hippocampal network activity (Hablitz, 1984). Future experiments using cell-type specific knockouts could better address the role of Kv2.1 in hippocampal circuitry (Wagnon et al., 2011).

Supplementary Material

Supplementary Material

Figure S1: Confirmation of C57BL/6J genetic background in Kv2.1^−/− line. The genetic backgrounds of two Kv2.1^+/− mice from our breeding colony were assessed at the DartMouse^™ Speed Congenic Core Facility at the Geisel School of Medicine at Dartmouth. DartMouse uses the Illumina, Inc. (San Diego, CA) GoldenGate Genotyping Assay to interrogate 1449 SNPs spread throughout the genome. The raw SNP data were analyzed using DartMouse’s SNaP-Map^™ and Map-Synth^™ software, allowing the determination for each mouse of the genetic background at each SNP location. This analysis indicates that the genetic background of this line is >97% C57BL/6J, with the remainder being 129Sv.

Figure S2: Kv2.1^+/+ and Kv2.1^−/− neurons yield similar K current after Kv2.1 inhibition. (a) Representative current traces from Kv2.1^+/+ and Kv2.1^−/− mouse hippocampal neurons (cell capacitance is 23.6pF and 17pF respectively). Voltage clamp command indicated below. Black traces, currents in BSA vehicle. Dark gray and light gray traces, currents in presence of 100 nM and 1000nM GxTX respectively. Arrow indicates the peak current used for analysis. Boxes indicate the regions of sustained and slow tail currents used for analysis. Peak (b), sustained (c), and tail currents (d) from Kv2.1^+/+ and Kv2.1^−/− neurons treated with BSA vehicle, 100nM GxTX, and 1000nM GxTX. Voltage stimulus same as panel A except step voltage set to −20mV (left panels) and 40mV (right panels). Black circles, Kv2.1^+/+. Red circles, Kv2.1^−/−. Thick bar is mean, error bars SEM.

Figure S3: Supporting data for field recordings. (a) Summary of evoked field potential responses from the CA1 region of hippocampal slices prepared from Kv2.1^+/+ and Kv2.1^−/− littermate mice before bicuculline application (“control” in Fig. 3a). Neither fEPSP duration nor amplitude differs between slices from the two genotypes (Kv2.1^+/+, n = 9 slices; Kv2.1^−/−, n = 8 slices). (b) Spontaneous epileptic discharges induced by bicuculline interfere with the development of LTP. Hippocampal slices from Kv2.1^+/+ were exposed to a TBS protocol (as described in the Methods) in the presence of 10μM bicuculline throughout the duration of the sixty minute recording period. There was no significant difference between the fEPSP slope before and after TBS (n=7 slices from 3 separate Kv2.1^+/+ animals). (c) LTP induced by TBS in Kv2.1^+/+ and Kv2.1^−/− slices in ACSF (3mM KCl). Upper traces of (c) show representative traces before TBS (control, gray line) and 55 min. after TBS (black line) from Kv2.1^+/+ and Kv2.1^−/− hippocampal slices. Bottom of (c) indicates the time course of normalized fEPSP slope as mean ± SEM (Kv2.1^+/+, n = 15 slices from 7 mice; Kv2.1^−/−, n = 6 slices from 3 mice). LTP at the CA3-CA1 synapse was induced by TBS at time point 0 (indicated by black arrow), and monitored for 60 min. after induction. Summary bar graph indicates that TBS induced LTP in both Kv2.1^+/+ and Kv2.1^−/− hippocampal slices; however, LTP amplitude was smaller in Kv2.1^−/− slices at 45–55 min after TBS. Kv2.1^+/+: 213%±9%, n=15 slices from 7 mice, *** P <0.0001, Student’s t-test; Kv2.1^−/−:180%±10%, n=6 slices from 3 mice, ****P<0.0001, Student’s t-test. Data are expressed as mean ± SEM. *P < 0.05.

Table S1: Incidence of status epilepticus and mortality in Kv2.1^−/− mice.