|Home | About | Journals | Submit | Contact Us | Français|
The somatodendritic A-current, ISA, in hippocampal CA1 pyramidal neurons regulates the processing of synaptic inputs and the amplitude of back propagating action potentials into the dendritic tree, as well as the action potential firing properties at the soma. In this study, we have used RNA interference and over-expression to show that expression of the Kv4.2 gene specifically regulates the ISA component of A-current in these neurons. In dissociated hippocampal pyramidal neuron cultures, or organotypic cultured CA1 pyramidal neurons, the expression level of Kv4.2 is such that the ISA channels are maintained in the population at a peak conductance of approximately 950 pS/pF. Suppression of Kv4.2 transcripts in hippocampal pyramidal neurons using an RNA interference vector suppresses ISA current by 60% in 2 days, similar to the effect of expressing dominant-negative Kv4 channel constructs. Increasing the expression of Kv4.2 in these neurons increases the level of ISA to 170% of the normal set point without altering the biophysical properties. Our results establish a specific role for native Kv4.2 transcripts in forming and maintaining ISA current at characteristic levels in hippocampal pyramidal neurons.
A-current is found throughout the somatodendritic compartment of pyramidal neurons and has clear functional importance in both the soma and dendrites of these cells. Dynamic regulation of the inactivated state is thought to provide an important associative signal that links patterns of electrical activity with changes in the excitability of a neuron (Schrader et al. 2002; Watanabe et al. 2002; Frick et al. 2004). In hippocampal CA1 pyramidal neurons, A-currents regulate the processing of synaptic inputs in the dendritic tree, the amplitude of back propagating action potentials into the dendritic tree, as well as the neuronal action potential firing properties at the soma (Hoffman et al. 1997; Ramakers and Storm 2002; Cai et al. 2004; Kim et al. 2005). Dynamic regulation of A-current levels produces excitability changes that can be seen at the level of a whole neuron, or within a small well-defined region of the dendritic tree (Ramakers and Storm 2002; Johnston et al. 2003; Cai et al. 2004; Frick et al. 2004; Kim et al. 2005).
A fundamental problem in neuroscience is the identification of native ion channels that are formed from specific genes in order to understand how regulation of these genes translates into changes in cellular electrophysiological properties. For our studies, we are interested in the expression and regulation of the ISA component of A-current (Jerng et al. 2004b). ISA is the classic, ‘subthreshold’ A-type current and is often referred to only as the A-current; however, biophysical studies in numerous cell types, including our studies here, clearly show that multiple A-type currents are found in individual neurons, and that the ISA current is a specific component of this overall A-current (Surmeier et al. 1989; Wu and Barish 1992; Mu et al. 1999; Chen and Johnston 2004).
Several previous studies have attempted to determine the underlying gene products responsible for forming A-currents in hippocampal pyramidal neurons (Ramakers and Storm 2002; Cai et al. 2004; Kim et al. 2005). These studies have shown that dominant -negative Kv4 constructs, as well as heteropodotoxins, which appear specific to Kv4 channels, suppress A-currents in these neurons. However, these previous studies, while offering important insights into the physiological and functional roles of Kv4 channels in pyramidal neurons, did not identify a specific gene product responsible for forming a specific component of A-current because of intrinsic limitations in the methods used in these studies. In this work, in order to overcome the limitations present in these previous studies on hippocampal pyramidal neurons, we have taken a novel approach to test if a specific gene, Kv4.2 (KCND2), is responsible for forming the ISA component of A-current in hippocampal pyramidal neurons. First, we have carefully separated the different A-current components in these neurons to determine if there is a specific effect of manipulating Kv4.2 only on the ISA component of the A-current. Previous studies in the heart have shown that Kv4 subunits are only responsible for forming one specific component of the A-current called Ito,f (Barry et al. 1998; Wickenden et al. 1999; Guo et al. 2005). In hippocampal neurons, previous dominant-negative and toxin studies have not clearly separated out the different current components forming A-current in these neurons to determine which components are selectively formed from Kv4 channels (Ramakers and Storm 2002; Cai et al. 2004; Kim et al. 2005). Second, in order to directly address the molecular link between a specific channel gene and a native ion channel type, we have turned to RNA interference (RNAi) technology as a means to acutely suppress a specific channel gene to probe the importance of that gene in forming the native channel (Brummelkamp et al. 2002; Couzin 2002). By down-regulating transcription of the Kv4.2 channel gene using RNA interference, and up-regulating the same transcript using over-expression vectors, we tested if changes in the expression of Kv4.2 are linked to changes in ISA current amplitude or biophysical properties in hippocampal pyramidal neurons. Previous studies using dominant negatives or toxins could only implicate Kv4 family members in the formation of A-type currents, but not a specific gene product. Our studies, presented here, provide strong direct experimental evidence for a specific role of the Kv4.2 gene in the formation of the ISA current in hippocampal pyramidal neurons, as well as a role for Kv4.2 transcript levels in regulating the amplitude of the specific ISA current component.
Our RNA interference vector, pSuperRed, was constructed by moving the shRNA expression cassette from the origin H1 variant of pSuper (Brummelkamp et al. 2002), into the pDsRed2-C1 vector (BD Biosciences Clontech, Palo Alto, CA, USA) which provides Kan/Neo selection along with CMV promoter diving expression of DsRed2 fluorescent proteins. We first removed BglII and HindIII sites from the pDsRed2-C1 vector polylinker to free these sites up for cloning of double-stranded oligos into the shRNA expression cassette. A 310-bp fragment from modified pDsRed2-C1 was removed by a full digestion with DraIII and a partial digestion with BspH1. An H1 promoter cassette for expression of shRNAs was prepared by cutting the H1 variant of pSuper with DraIII and AflIII. This fragment was cloned into the modified pDsRed2-C1 to create pSuperRed, see Fig. 4. Interfering RNAs were cloned into the BglII and HindIII sites that follow the H1 promoter as shRNAs. Kv4.2 was screened for likely RNAi target sequences using the criteria of Brummelkamp et al. (2002). Two target sequences were identified in the rat Kv4.2 sequence that corresponded to a AA(N19)TTconsensus with 30–70% GC content. The most effective sequence, used to make pSuperRed-Kv4.2i, is AAAGGACTCAGGACGCTCTAATT which targets the CDS of rKv4.2 at residues 106–128 (shown in bold). The shRNA insert for pSuperRed-Kv4.2i was made by cloning the following sequence: CCCAGGACTCAGGACGCTCTAA ttcaagagaTTAGAGCGTCCTGAGTCCTTTTTTGGAAA into BglII + HindIII cut pSuperRed by ligation of double-strand oligonucleotides with GATC and AGCT sticky ends. Our dominant-negative Kv4 construct, Kv4DN-enhanced green fluorescent protein (EGFP), was a kind gift of David Johns (Johns Hopkins, Baltimore, MD, USA), and has been described and characterized previously as GFP-Kv4.2ST (Johns et al. 1997). This construct acts as a dominant negative because EGFP is cloned into the Pore region of Kv4.2, thus creating a truncated dominant-negative subunit, that retains the T1 assembly domain, but lacks the required pore-forming sequence to create functional channels. Kv4DN-EGFP was cloned into pSinRep5 (Invitrogen, Carlsbad, CA, USA) to make pSindbis-Kv4DN-EGFP. A dual subgenomic vector co-expressing EGFP with Kv4.2 was also made (Green-Sindbis-Kv4.2). Recombinant Sindbis viruses were generated as described previously (Jeromin et al. 2003). Briefly, RNA was transcribed in vitro from the pSindbis vectors and the DH-26S helper plasmid using the Megascript kit (Ambion, Austin, TX, USA). Sindbis and DH-26S RNAs were co-electroporated into BHK cells (ATCC, Manassas, VA, USA) using either an Invitrogen electroporator or the Bio-Rad II gene pulser (Bio-Rad Laboratories, Hercules, CA, USA). The use of the DH-26S helper ensured the preferential infection of neurons in slices. Supernatants containing the recombinant virus were harvested 48–72 h after electroporation, and the titer of the harvested viruses was typically in the range of 108–109/mL.
COS7 cells (ATCC) were grown in Dulbecco’s modified Eagle’s medium plus 10% fetal bovine serum (HyClone, Logan, UT, USA) and 1% penicillin/streptomycin (Invitrogen). Cells were plated at a density of 150 000 cells per 35-mm dish, grown overnight and transfected the next day with Fugene6 (Roche Molecular Biochemicals, Indianapolis, IN, USA) following the manufacturer’s protocols. To test pSuperRed constructs, cells were co-transfected with the pSuperRed-RNAi vector and a pCMV-target expressing vector. After 48 h, the cells were harvested and the expression level for the target protein was assayed by western blotting.
All animal procedures were conducted in accordance with the policies of AALAC and Baylor College of Medicine IACUC. Methods for dissociated culture of CA1-CA3 neurons from 0- to 3-day-old Sprague–Dawley rats were adapted from Malgaroli and Tsien (1992). Glass coverslips (12 mm round) were placed in 24-well plates and coated with a 1/20–1/50 solution of Matrigel (BD Biosciences Clontech) and growth medium immediately prior to the neuron dissociation protocol. Post-natal day 1-2 rat pups (Sprague–Dawley; Harlan Inc., Indianapolis, IN, USA) were killed by decapitation and the brains were immediately placed in divalent-free Hank’s salt solution (Sigma, St Louis, MO, USA) plus 20% fetal calf serum (HyClone). Hippocampi were isolated from the rest of the brain and, in subsequent steps, particular attention was paid to removal of the entorhinal cortex and subiculum, as well as all visible vasculature. The hippocampi were then placed in fresh Hank’s salt solution + 20% fetal calf serum medium and crudely dissociated using a scalpel. Hippocampi were digested at 37 °C for 10 min in a digestion solution containing 10 mg/mL trypsin (Sigma), washed, and then mechanically triturated in dissociation solution containing Hank’s salt solution plus MgSO4 using siliconized pipettes. After dissociation, cells were spun at 1000 g for 10 min, re-suspended in growth medium, and plated at a density of 2000–5000 cells per well. One litre of basic culture medium consisted of phenol-free minimal essential medium (MEM; Invitrogen) supplemented with 5 g glucose (Sigma), 200 mg NaHCO3 (Sigma), and 100 mg Transferrin (Calbiochem, La Jolla, CA, USA). Growth medium consisted of basic culture medium plus 5% fetal bovine serum (HyClone), 2% B-27 supplement (Invitrogen), 500 nm l-glutamine (Sigma), and 2–4 μm cytosine-D-arabinofuranoside (Sigma). For some experiments, 0.25 g/L insulin (Sigma) was also included in the culture media. Cultures were maintained at 37 °C in a 5% CO2-humidified incubator for 7–14 days in growth medium prior to transfection. Transfection was performed using either Lipofectamine 2000 (Invitrogen) or Ca-phosphate precipitation. For Lipofectamine 2000, growth medium was removed and cultures were washed once with serum-free medium. For each well, a 100-μL mixture of Lipofectamine 2000 (2 μL), DNA (1 μL), and serum-free medium (97 μL) was added directly to the coverslip in a drop-wise manner. Subsequently, 300 lL of growth medium was added to each well and cultures were incubated for 24–48 h before analysis. For Ca-phosphate transfections, neuronal cultures were washed twice with fresh, serum-free medium (MEM + 5 g/L glucose) after saving the original media. A total of 2 μg DNA was mixed with 2 μL 2 m CaCl2 and distilled water in a volume of 15 μL, and then an equal volume of 2 × HEPES-buffered saline (BD Biosciences Clontech) was added to a total volume of 30 lL. After 5 min at 22 °C, the mixed solution was added drop-wise to each coverslip, and the cells were returned to the incubator (37 °C) for 30–45min. To minimize cytotoxicity and optimize transfection efficiency, the precipitate was closely monitored; when a layer of precipitate became obvious, cells were washed three times with MEM and returned to the saved original medium.
Organotypic slices were prepared and cultured according to the interface technique as originally described by Stoppini et al. (1991) and Jeromin et al. (2003). Infection of neurons with Sindbis virus was accomplished by focal injection in CA1 pyramidal neurons in organotypic slices or application to primary cultures of hippocampal neurons (Jeromin et al. 2003).
Neurons were visualized using upright microscopes equipped with IR-DIC and epifluorescence (either an Olympus BX51 upright microscope with a 60 × water immersion objective or a Zeiss Axioskop with a 40 × water-immersion objective). Voltage-clamp protocols were conducted in whole-cell configuration using Axo-patch 200b amplifiers. Pulse generation and data acquisition were controlled with either custom software written for the Igor Pro analysis environment (Yuan et al. 2002), or pClamp7–9 software. Whole-cell capacitance and series resistances were compensated to more than 80%, and series resistances were less than two times the tip resistance. Leakage and capacitive currents were digitally subtracted online. Current inactivation kinetics was determined by single exponential curve fitting. Significance (p < 0.05) was determined by two-sample t-tests. Error bars represent SE. Voltage-clamp protocols to isolate current components are described in the text.
Neurons were recorded in a perfusion chamber at 22 °C for all experiments. To avoid variations in native ISA density, and to reduce the complexity of the molecular identity of native ISA among different cell types, we chose to restrict our studies to a subset of transfected pyramidal-shaped neurons with a major apical dendrite, some basal dendrites, and a relatively small soma with a whole-cell capacitance around 20 pF. Bath solution was continuously bubbled with a 95%/5% mixture of O2/CO2 and consisted of, in mm:NaCl 125, KCl 2.5, NaHCO3 25, CaCl2 2, MgCl2 2, dextrose 10. Tetrodotoxin (Tocris, Ellisville, MO, USA) was added to the bath solution prior to experiments at 1–2-μm concentrations. Heteropodatoxin-3 (HpTx3; NPS Pharmaceuticals, Salt Lake City, UT, USA) was dissolved in water and added to the external solution with 0.1% bovine serum albumin. Capillaries (1.5 mm) were pulled to resistances of between 3 and 6 MΩ and were filled with internal solution consisting of, in mm: Kgluconote 120, KCl 20, NaCl 5, HEPES 10, Mg-ATP 4, Tris-GTP 3, phosphocreatine 14. In some experiments, to further isolate Kv currents, 2 mm MnCl2 was substituted for CaCl2 in the bath to block Ca2+ currents, and Ca2+-activated K+ currents. Activation and inactivation data analyzed as both fits to averaged data (presented in the figures), and averaged data for the fits to individual experiments, used to determine mean and standard error. Numbers (n) of 7–16 were used in these experiments. There were no significant differences in the average activation or inactivation gating properties of ISA, IAD or ISS with knockdown or over-expression of Kv4.2.
For isolation of ISA, we subtracted recordings performed at test potentials from −70 to +50 mV, obtained by holding, or pre-pulsing from 200 ms to 1 s, at −30 to −40 mV from matched recordings obtained by pre-pulsing or holding at more negative potentials, typically −80 to −100 mV. The subtraction selectively removes the IAD and ISS components from the recording. Note that this protocol produces a small subtraction artifact in the pre-pulse as a result of the partial activation of ISS currents at −30 mV. This artifact causes the pre-pulse baseline current to jump around in Fig. 5(c) and Fig. 6(c). For IAD, we subtracted recordings performed at test potentials from −40 to +90 mV obtained by pre-pulsing for 500 ms to 1 s, at +50 mV, then stepping to −30 mV for 10 ms, from recordings obtained by holding or pre-pulsing for 500 ms to 1 s at −30 mV. This procedure subtracts ISS from combined IAD and ISS recordings. The brief step to −30 mV was carried out to obtain more accurate subtractions by closing ISS channels before the subtraction. For ISS, we held at +50 ms for 500 to 1 s to inactivate all A-type channels, stepped to −30 mV for 10 ms to close ISS channels, then applied a series of activation pulses to record an isolated ISS current. To quantify the magnitude of three current components in the same recordings, we constructed inactivation curves as shown in Fig. 1(b), then fit these curves for the two Boltzmann components and the residual non-inactivating component. The ISA amplitude was obtained by the amplitude of the most negative Boltzmann component, the IAD amplitude was obtained by the amplitude of the less negative component, and the ISS component was obtained by the residual component. We also separately verified that the amplitudes of these current components measured from the inactivation curve closely matched the amplitudes of the isolated currents recorded as described above.
Whole cell lysates were separated by sodium dodecyl sulfate – polyacrylamide gel electrophoresis and transferred to Immobilon filters (Millipore Corporation, Bedford, MA, USA). Every gel contained both a control and experimental condition transfected in triplicate, and was probed either simultaneously or sequentially for the protein of interest. Glyceraldehyde 3-phosphate dehydrogenase (GAPDH) levels were also probed by western blotting to serve as an internal control. The following primary antibodies were used: anti-kv4.2 (rabbit, AB no. 5362; Chemicon, Temecula, CA, USA), anti-Kv4.3 (goat, n15; Santa Cruz Biotechnology, Santa Cruz, CA, USA), anti-Kv1.4 (rabbit, Ab5926; Chemicon), anti-KChIP3 (rab-bit, FL-214; Santa Cruz Biotechnology), and anti-GAPDH (mouse, clone 6c5; Advanced Immunochemicals, Long Beach, CA, USA). Horseraddish peroxidase-conjugated secondary antibodies (Pierce, Rockford, IL, USA) were used for secondary antibody detection on Kodak MR Film. Films were scanned into a computer and luminance densitometry was performed using Optiquant software (Perkin Elmer, Wellesley, MA, USA).
Organotypic hippocampal slices were fixed by immersion with 3.75% acrolein in a solution of 2% paraformaldehyde and 0.1 m phosphate buffer (PB), pH 7.4, for 1 h. The slices were then post-fixed for 1–2 h in 2% paraformaldehyde. To remove the excess of reactive aldehydes, sections were treated with 1% sodium borohydride in 0.1 m PB for 30 min. Sections were then rinsed with 0.1 m PB followed by 0.1 m Tris-buffered saline, pH 7.6. Hippocampal slices were processed for immunoperoxidase and immunogold-silver labeling of Kv4DN-EGFP prior to plastic embedding using an affinity-purified rabbit polyclonal antibody directed against highly purified recombinant EGFP made in Escherichia coli (Abcam Ltd, Cambridge, UK). The selectivity of this antibody has been established by western blotting, immunoprecipitation and immunofluorescence experiments (Jasinska et al. 1999; McCabe and Berthiaume 1999). In addition, no labeling was seen in uninfected neurons or when the antibody was omitted from the reaction procedure. For this experiment, the slices were first incubated for 30 min in 0.5% bovine serum albumin in Tris-buffered saline to minimize non-specific labeling. The slices then were incubated for 48 h at 4 °C in 0.1% bovine serum albumin in Tris-buffered saline containing rabbit polyclonal antiserum for EGFP (1: 500 dilution for immunoperoxidase, 1: 100 dilution for immunogold). Subsequently, the EGFP antibody was detected by using the pre-embedding immunoperoxidase or immunogold-silver methods. Immunolabeled hippocampal slices were rinsed in 0.1 m PB pH 7.4, and then post-fixed for 1 h in 2% osmium tetroxide in PB. They were subsequently dehydrated through a graded series of ethanols and propylene oxide prior to embedding in Epon 812 (Shell Chemicals, London, UK) between sheets of Aclar plastic (Honeywell, Morristown, NJ, USA). Ultrathin sections were cut with a diamond knife (Diatome, Fort Washington, PA, USA), and collected on copper mesh grids. The sections on grids were counterstained with uranyl acetate and lead citrate, and examined with a JEOL 1010 electron microscope (Reynolds 1963).
We first developed biophysical protocols to separate the major components of inactivating current in our hippocampal neuronal cultures to determine if we could isolate a clear ISA component from other voltage-gated potassium currents. Hippocampal neuronal cultures from post-natal day 2 rats were prepared and cultured for 9–14 days. Pyramidal-shaped neurons with capacitances around 20–25 pF were selected for our whole-cell voltage-clamp recordings. A steady-state inactivation protocol was applied by holding neurons at −80 mV and then pre-pulsing to potentials between −100 mV and +40 mV, before applying a test pulse to +50 mV. As can be seen in the recordings from a representative cell, a pre-pulse potential to −100 mV reveals a large component of inactivating current that is progressively lost at more positive pre-pulse potentials (Fig. 1a). Interestingly, at a middle potential range in the −40 to −30 mV range, the inactivation curve flattens out before becoming steeper again around −10 mV, and then flattens out again with a substantial component of non-inactivating current. When peak current is plotted versus potential, two clear components of inactivating current can be seen (Fig. 1b). Double Boltzman fits show that the more negative inactivating component has an inactivation midpoint of −70.1 ± 2.2 mV, as expected for ISA. The more positive inactivating component has a midpoint of −11.4 ± 2.3 mV, and seems to correspond well to a ‘d-type’ inactivating channel described in cultured hippocampal pyramidal neurons and in patch recordings from rat hippocampal dendrites (Mu et al. 1999; Chen and Johnston 2004). We will refer to this as the IAD current. Finally, the non-inactivating voltage-gated K current, likely composed of M-channels and delayed rectifier channels, will be referred to as the sustained current, ISS (Gu et al. 2005; Misonou et al. 2005). ISA can easily be separated from the other components by using a difference current measurement between traces recorded with pre-pulses to −100 mV versus −30 mV. As can be seen in Fig. 1(c), subtraction of the −30 mV traces from the −100 mV traces reveals the rapidly inactivating ISA current. Using this isolation protocol, we characterized the peak activation properties of ISA in our neurons and plotted the peak conductance probability versus potential in Fig. 2(a). The measured midpoint for peak activation of this channel is −22.1 ± 2.9 mV. If neurons are held at more positive potentials, we can also characterize the activation properties of IAD and ISS. To isolate IAD, a difference current method was applied. Voltage steps from a holding potential of −30 mV activate both IAD and ISS. Applying a pre-pulse to +40 mV inactivates IAD, allowing us to selectively record ISS. Because this pre-pulse potential is large enough to activate ISS, we applied a brief pulse to −30 mV to shut the ISS channels before applying the test pulses. The ISS current activation curve obtained in this manner has a half activation point of 2.8 ± 2.7 mV (Fig. 2c). Subtraction of the ISS current from the current recorded from a holding potential of −30 mV reveals the more slowly inactivating IAD current with an inactivation time constant of 117 ± 18 ms at +50 mV (Fig. 2b). The activation curve obtained from this subtraction suggests a half activation point of +28.9 ± 3.3 mV for IAD.
We next sought to determine if the current we have identified as ISA is sensitive to the toxin heteropodatoxin-3, HpTx3. In heterologous expression systems, heteropodatoxins have been shown to selectively suppress the Kv4 class of potassium channel (Sanguinetti et al. 1997; Bernard et al. 2000; Zarayskiy et al. 2005). Previous studies on CA1 pyramidal neurons in acute slices have shown that at least part of the A-current in these neurons is sensitive to heteropodatoxin (Ramakers and Storm 2002). The toxins appear to act by modulating the gating properties of Kv4 channels, producing a strong suppression of Kv4 channel currents at negative potentials that is relieved at more positive potentials (Sanguinetti et al. 1997; Zarayskiy et al. 2005). Whole-cell voltage-clamp recordings were obtained on hippocampal pyramidal neurons in culture and sensitivity of currents was tested by application of 100 nm HpTx3. As can be seen in Fig. 3(a), 100 nm HpTx3 produces a selective reduction in the fast inactivating ISA component of current with no effects on the slower inactivating IAD or sustained ISS components. We next compared the ability of HpTx3 to suppress ISA current at different test potentials (Fig. 3b). The results show a clear voltage dependence to HpTx3 suppression of ISA, with a near 100% block at −40 mV reduced to a 40% block at 0 mV, similar to the results previously described on Kv4.2 in heterologous systems (Sanguinetti et al. 1997; Zarayskiy et al. 2005).
Previous studies have suggested that the ISA current in a variety of neurons is likely made from Kv4 family proteins (Baldwin et al. 1991; Sheng et al. 1992; Serodio et al. 1996; Jan and January 1997; Song et al. 1998; Tkatch et al. 2000; Jerng et al. 2004b). In situ hybridization and immunostaining have shown that Kv4.2 is expressed in hippocampal pyramidal neurons, making it a logical candidate gene for formation of ISA in our neurons (Serodio et al. 1996; Rhodes et al. 2004). To directly test this hypothesis, we constructed an RNA interference vector that allows us to knock down specific mRNAs, while marking transfected neurons with a red fluorescent protein, by cloning the H1 promoter region from pSuper into pDsRed2-C1 (Clontech) to make pSuperRed, see Fig. 4(a and b). RNAi targets on Kv4.2 were identified and screened by blast searching to confirm that they are uniquely present on Kv4.2. To express RNAi, shRNAs containing the RNAi inserts were designed as 64-base complementary DNA oligonucleotide pairs (Sigma Genosys, Houston, TX, USA) with 5′ overhangs matching the BglII and HindIII restriction sites that follow the H1 promoter on pSuperRed. Constructs were tested by co-transfection of pSuperRed with pCMV-Kv4.2 into COS7 cells and western blotting. An effective RNAi sequence was identified that corresponds to a 19-bp sequence 108–126 bases downstream of the rat Kv4.2 start codon to make pSuperRed-Kv4.2i. Following co-transfection of COS7 cells with pCMV-Kv4.2 plus pSuperRed or pSuperRed-Kv4.2i, the expression of Kv4.2 protein was dramatically reduced only in the presence of pSuperRed-Kv4.2i (Fig. 4c). We confirmed the specificity of pSuperRed-Kv4.2i by testing if expression of other proteins that might contribute to forming transient potassium currents in hippocampal pyramidal neurons were affected by pSuperRed-Kv4.2i. As can be seen in Fig. 4(d), following co-transfection of pSuperRed-Kv4.2i with a series of potential target constructs in COS7 cells, pSuperRed-Kv4.2i specifically knocks down the expression of Kv4.2 by over 80%, but has no effect on the expression of Kv4.3, Kv1.4, or KChIP3. In addition, the endogenous expression of GAPDH is not affected by pSuperRed-Kv4.2i. As expected, the control vector pSuperRed has no effect on expression of any of these proteins, showing that the selective knockdown of Kv4.2 by pSuperRed-Kv4.2i is specifically produced by the inserted RNAi sequence.
Our recordings show that an ISA current that is sensitive to HpTx3 is present in whole-cell recordings from cultured rat hippocampal pyramidal neurons. Therefore we next sought to test if Kv4.2 is necessary for expression of ISA in these neurons by suppression of Kv4.2 using RNA interference. Hippocampal neurons in culture were transfected with pSuperRed or pSuperRed-Kv4.2i and examined 48 h after transfection. Approximately 1–5% of the neurons in these cultures were transfected under these conditions. Red fluorescent neurons with pyramidal shapes were selected for voltage-clamp recordings, see Fig. 4(b). Current components were isolated using the differential inactivation properties described previously. The properties and the amplitudes of these current components were compared with and without the expression of Kv4.2 RNAi. A consistent finding in these studies was that the isolated ISA current component is reduced in neurons transfected with pSuperRed-Kv4.2i, see Fig. 5(a). Separation of all three current components and analysis of the average amplitude data (Fig. 5b), shows the effects of Kv4.2 RNAi are specific to the reduction of ISA current amplitude by an average of approximately 60%. The measured ISA current in neurons expressing pSuperRed-Kv4.2i is significantly different from both untransfected neurons or neurons expressing pSuperRed alone. There is no significant difference in ISA current amplitude produced by the control pSuperRed vector. The data also show that there are no significant differences in IAD or ISS current amplitudes between control neurons, pSuperRed transfected neurons or neurons transfected with pSuperRed-Kv4.2i.
The primary effects of Kv4.2 RNAi on ISA are only on the amplitude of the ISA current. No significant differences were observed in the activation or inactivation properties of any of the current components following transfection with pSuperRed-Kv4.2i. Scaled ISA current traces in response to a pulse to +40 mV show that the residual ISA current after Kv4.2 RNAi is kinetically similar to the overall ISA current, see Fig. 5(c), as are the steady-state activation and inactivation curves (Fig. 5d). The inactivation time courses for the normal ISA current and residual ISA compared across a range of voltages in the two conditions are not significantly different (Fig. 5e).
Expression of Kv4.2 RNAi in hippocampal pyramidal neurons shows that ISA current is very sensitive to the suppression of Kv4.2 expression. Next, we wished to test if increases in the expression of Kv4.2 can specifically increase the amplitude of ISA to determine if Kv4.2 message levels control the amplitude of ISA current. Neurons were co-transfected with pSuperRed along with pCMV-Kv4.2 and compared 48 h after transfection with neurons transfected with pSuperRed alone. Red fluorescent neurons with a clear pyramidal shape were selected for voltage-clamp studies to determine if the increase in Kv4.2 expression specifically modulates a single component of voltage-gated potassium current. The current components were isolated as described previously and compared. Increasing the expression of Kv4.2 had a clear specific effect to increase the amplitude of the ISA component with no significant changes in the IAD or ISS currents (Figs 6a and b). We further characterized the biophysical properties of the isolated ISA currents under the normal and over-expressing conditions. The results were strikingly similar in the two cases. When scaled to the same size, the ISA currents in neurons over-expressing Kv4.2 have a similar shape to those from control neurons (Fig. 6c). In addition, the time constants and voltage dependence for activation and inactivation are similar between control and Kv4.2 over-expressing neurons, see Fig. 6(d and e). These results strongly support our hypothesis that Kv4.2 message levels control the amplitude of the ISA current in hippocampal pyramidal neurons.
We next examined the rate of recovery from inactivation for transient potassium channels with and without changes in Kv4.2 transcription (Fig. 7a). Recovery from inactivation was characterized using a standard two-pulse protocol, where the time of the interpulse potential at −90 mV was varied. The amplitude of the current in the second pulse was compared with the first pulse and plotted versus the time at −90 mV. Control neurons show two clear components of recovery from inactivation at −90 mV, one with a time constant of 34.1 ± 2.4 ms and a second slower recovering component with a time constant of 542 ± 123 ms. Recovery traces were recorded and compared between control neurons expressing pSuperRed with those expressing either pSuperRed-Kv4.2i or co-expressing pCMV-Kv4.2 with pSuperRed. The plots show that the time constants of the two components in the different conditions were not significantly different, but the amplitudes are significantly changed (Fig. 7b). In the presence of pSuperRed-Kv4.2i, the more rapidly recovering component is smaller in amplitude compared with the slower component, and the opposite is true with the increase in expression of Kv4.2 using pCMV-Kv4.2 (%Fast amp: pSuperRed, 74 ± 3%; pSuperRed-Kv4.2i, 53 ± 5%; pCMV-Kv4.2, 87 ± 2%). These results suggest that the faster component is dependent upon the Kv4.2 expression level for Kv4.2. To confirm this result and link the faster component to ISA and the slower component to IAD, we compared the amplitudes of the isolated current components in these neurons to the amplitudes of the fast and slow recovery components. In Fig. 7(c), we plot the summary data comparing the fraction of total transient current that is ISA with the fraction of the total recovery that occurs with the fast kinetic for neurons expressing pCMV-Kv4.2, pSuperRed, and pSuperRed-Kv4.2i. The identity line is drawn to show the expected result if the two components are identical. The results closely match the predicted trend line, supporting an identity of the fast component matching recovery of the ISA channels and the slower component as the IAD recovery that is not sensitive to Kv4.2 expression levels.
We next sought to compare the effects of dominant-negative Kv4 constructs, that suppress the functional expression of all Kv4 channels, to the specific effects of suppressing Kv4.2 with RNA interference. A GFP-tagged dominant negative Kv4 subunit (Kv4DN-EGFP) was expressed in our cultured hippocampal pyramidal neurons using a CMV promoter vector (pCMV-Kv4DN-EGFP). Currents in neurons expressing pCMV-Kv4DN-EGFP were compared with neurons expressing pCMV-EGFP. Similar to our results with Kv4.2 RNAi, and in agreement with results from Kv4 dominant-negative constructs expressed in other systems, we see a selective reduction in ISA current amplitude following expression of Kv4DN-EGFP compared with control (Cai et al. 2004; Kim et al. 2005). The magnitude of the suppression of ISA with pCMV-Kv4DN-EGFP was compared with the suppression seen with pSuperRed-Kv4.2i (Fig. 8a). Both constructs produced an approximate 60% suppression of ISA by 48 h, significantly different from control. We do not detect a significant difference in the magnitude of the current that is suppressed when comparing between Kv4DN-EGFP and Kv4.2 RNAi.
To further confirm that the dominant-negative Kv4 subunits are suppressing the same component of current as Kv4.2 RNAi, we characterized the effects of Kv4DN-EGFP on the recovery from inactivation. Neurons were transfected with pCMV-EGFP or pCMV-Kv4DN-EGFP and recovery from inactivation characterized using a two-pulse protocol as described above, see Fig. 8(b). The magnitude of the transient current in the second pulse relative to the first pulse was plotted versus the recovery time in Fig. 8(c). As described previously, the recovery follows a double exponential time course. There was no significant difference in the time constants for recovery when comparing control with Kv4DN-EGFP; however, the amplitude of the two components was significantly changed, see Fig. 8(d). The relative amplitude of the fast component was reduced compared with the slower component following expression of Kv4DN-EGFP. These results are again in good agreement with our studies using Kv4.2 RNAi, where a selective change in the recovery of the fast component was linked to the magnitude of change in the ISA current, see Fig. 7(d). We therefore conclude that Kv4.2 RNA interference and Kv4DN-EGFP both selectively suppress the ISA current in cultured hippocampal pyramidal neurons to a similar extent.
Our studies show that the amplitude of the ISA current in hippocampal pyramidal neurons in culture is dramatically modulated by manipulations that increase or decrease the levels of Kv4.2 transcripts. Next, we wanted to extend these results to a specific neuronal population, CA1 pyramidal neurons. To accomplish this, we switched to organotypic slice cultures of hippocampus and the use of Sindbis viral vectors to modulate ISA currents (Jeromin et al. 2003). We developed two different Sindbis vectors to modulate ISA currents: (i) Sindbis-Kv4DN-EGFP which expresses the same EGFP tagged dominant negative Kv4 constructs; and (ii) a dual SGP promoter Sindbis vector that expresses EGFP from one promoter and Kv4.2 from another promoter (Green-Sindbis-Kv4.2). These two constructs closely mimic the CMV promoter constructs we used in dissociated culture and should allow us to compare the results from these two systems.
Neurons were infected in organotypic culture by focal injection of viral stocks into the CA1 pyramidal neuron layer of the cultured slice. Within 24 h, green fluorescent neurons could be detected that remained healthy by visual inspection and basic physiological parameters for the duration of the experiments (data not shown, see also Jeromin et al. 2003). Visually these vectors produce different patterns of green fluorescence when infected in neurons. When infected with Green-Sindbis-Kv4.2, the entire body and processes of the neuron fill completely with green fluorescence, Fig. 9(a). In contrast, the Sindbis-Kv4DN-EGFP construct shows a punctuate distribution of green fluorescence that is restricted to the somatodendritic compartment (Fig. 9b).
Previous studies in the heart have suggested that long-term exposure to dominant negative Kv4 constructs can produce structural changes in the heart that are likely unrelated to the specific effects of these constructs on Ito,f (Wickenden et al. 1999; Guo et al. 2005). To rule out obvious cytological effects in neurons during short-term exposure to these dominant negative viral constructs, and to further examine the subcellular basis for the observed punctate protein distribution, we performed immuno-electron microscopy on CA1 pyramidal neurons expressing Sindbis-Kv4DN-EGFP, stained with anti-EGFP antibodies. Following labeling with this specific antibody, electron microscopy revealed that only a subset of neurons in fields infected with our viral vector constructs showed the specific immuno-EM signals. Immunoperoxidase staining (open arrows) of the soma of infected neurons shows extensive labeling of internal membranes of the endoplasmic reticulum (ER) and Golgi stacks of neurons expressing Kv4DN-EGFP (Fig. 9c). No labeling is seen in uninfected neurons, or in terminals from uninfected neuron terminals (UT) that abut the soma of the infected neuron. Analysis of the structure of infected neurons revealed that there were no obvious cytological differences between labeled and unlabeled neurons in the same slices. Our analysis of the basis for the punctate distribution of this protein in dendrites (Fig. 9d) suggests that most of the puncta (open arrows) are associated with collections of protein in ER and multivessicular bodies and not with a specific incorporation of these constructs into asymmetric synapses (filled arrows). To quantitate the subcellular distribution of protein within the neuron, immunogold detection was performed (Fig. 9e). Of the 78 gold particles analyzed in the dendrites, only 12 (15%) were associated with surface membranes. The remaining particles were in cytosolic compartments often associated with ER or transport vesicles. In CHO cells, Kv4DN-EGFP also shows a concentration of labeling of intracellular membranes, in agreement with our results here (Johns et al. 1997). We therefore conclude that short-term use of such dominant negative viral constructs does not produce disruption of normal cytology and produces proteins that are largely found in internal membranes of the somatodendritic compartment.
Next, CA1 pyramidal neurons infected with Sindbis vectors were recorded by whole-cell patch clamp to compare the functional effects of these manipulations with our studies on dissociated hippocampal neurons in culture. We first compared the resting potential of control and infected neurons and found that there were no significant differences (uninfected: Vm = −61 ± 1.3 mV, n = 13; infected: Vm = −61.6 ± 1.2 mV, n = 15), further supporting electron microscopy picture that these neurons are essentially healthy 2 days following infection with these vectors.
We next wanted to compare the ISA currents recorded in identified CA1 pyramidal neurons in slice culture with those from pyramidal neurons recorded in dissociated culture to compare the channel densities, basic gating properties and importance of Kv4 channels for this current. No significant differences were noted in steady-state gating properties or kinetics of ISA currents in these two systems. Peak conductance density for ISA recorded in dissociated cultured hippocampal pyramidal neurons or CA1 pyramidal neurons in organotypic culture were also not significantly different, and averaged around 950 pS/pF, see Fig. 10(a). Given an estimated conductance for ISA channels of 6 pS, this corresponds to roughly 1.5–1.7 channels per square micron open at the peak conductance under both conditions (Chen and Johnston 2004).
We next examined the specific effects of these viral vectors on the ISA currents in infected CA1 neurons in organotypic culture. In agreement with our studies in dissociated culture, expression of Kv4DN-EGFP and Kv4.2 in CA1 pyramidal neurons selectively suppresses the levels of ISA current without affecting other voltage-gated potassium currents. Kv4DN-EGFP suppressed ISA by 60%, an amount similar to what we had previously seen in dissociated neuronal culture (Fig. 10b). Increasing transcription of Kv4.2 in CA1 pyramidal neurons increases the levels of ISA current to 175% of its original level, similar to what we have seen in dissociated cultures (Fig. 10d). Comparison of these data across groups shows that there are no significant differences in the effects on ISA current amplitude between expression of Kv4DN-EGFP in dissociated neuronal culture and CA1 neurons in organotypic slice culture (Fig. 10c). Likewise, the amplitude increase produced by over-expression of Kv4.2 is not significantly different between these two systems (Fig. 10e). We further compared the biophysical properties between these ISA currents recorded under these various conditions and we did not observe any significant differences in ISA channel function. We therefore conclude that ISA currents in hippocampal pyramidal neuron recorded in dissociated culture under our conditions largely reflect the current density and regulatory properties of CA1 neurons in hippocampal neuron in organotypic slice culture.
Using a novel approach combining careful biophysical separation of currents with RNA interference and over-expression, we have identified Kv4.2 (KCND2) as a critical gene in the formation and regulation of ISA current amplitude in hippocampal pyramidal neurons. Our studies in hippocampal pyramidal neurons in dissociated culture and CA1 pyramidal neurons in organotypic culture have shown that a distinct A-current component, the ISA current, is tightly regulated by changes in Kv4.2 channel transcription. Suppression of Kv4.2 expression using RNAi suppresses ISA, whereas increasing the level of Kv4.2 transcripts increases the amplitude of ISA. These studies therefore clearly identify Kv4.2 as a critical gene in the formation of hippocampal pyramidal ISA current. Our results also suggest a likely dynamic aspect to this regulation, with a role for the Kv4.2 gene in regulating the ISA current amplitude through changes in gene transcription. Previous in situ hybridization and immunohistochemistry studies have shown that the Kv4.2 gene is highly expressed in hippocampal pyramidal neurons; therefore, our studies directly link these specific gene products, present in these neurons, to the selective formation of the ISA component of A-current (Serodio et al. 1996; Rhodes et al. 2004).
The ISA current in cultured hippocampal pyramidal neurons formed from Kv4.2 subunits has a half activation potential of −22.1 ± 2.9 mV and a half inactivation potential of −70.1 ± 2.2 mV. This very negative half inactivation potential is considerably more hyperpolarized than the typical resting potential we record in these neurons, around −61 ± 2.9 mV. From our steady-state inactivation curve for ISA (Fig. 2a), this means that more than two-thirds of the ISA channels in these neurons are typically inactivated at rest. In contrast, the IAD current has a half inactivation potential of −11.4 ± 2.3 mV (Fig. 2b), and therefore is likely not inactivated at all at rest. The activation of the IAD current with a midpoint of +28.9 ± 3.3 mV requires much larger depolarizations than ISA. Therefore, it is clear that in native neurons the molecular compositions and functional ranges of these two A-type channel types are distinct.
Although Kv4.2 RNA interference selectively suppresses 60% of the ISA at 48 h, the identity of the residual unsuppressed ISA current component is unclear. Biophysical characterization of this residual ISA current suggests that the channels composing this current are very similar to the Kv4.2-sensitive current, making it likely that Kv4 family members are responsible for this current. In addition to Kv4.2, Kv4.3 is also expressed in the hippocampus, although at higher levels in interneurons, therefore it may be responsible for this residual current component (Rhodes et al. 2004). Interestingly, dominant-negative constructs that suppress all Kv4 channels suppress the ISA current to a similar extent as Kv4.2 RNA interference. Although neither RNA interference nor dominant-negative constructs are expected to suppress current by 100%, the simplest explanation for this common result is Kv4.2 is required for all ISA current expression and that 40% unsuppressed ISA current reflects channels that are not turned over in 48 h. If this is so, then Kv4.3 may not efficiently form ISA channels on it own in pyramidal neurons, similar to the heart where a Kv4.2 knockout mouse was shown to lack Ito,f despite the fact that Kv4.3 expression still remains, suggesting that Kv4.2 has a special role in the surface expression of function Ito,f channels (Guo et al. 2005). However, we must consider the fact that the apparent similarity in the effects of RNA interference and dominant-negative constructs may be a fortuitous result. In our future studies, we will examine the time course of Kv4.2 protein turnover in neurons, and removal from the cell surface, to determine if the 40% residual current is a result of either the slow turnover of all channels, or the presence of a more stable fraction of channels that turns over slowly. In addition, we are developing lentiviral vectors that should allow us to examine the currents in infected neurons expressing Kv4.2-RNAi or Kv4DN-EGFP at longer time points, as well as vectors that can suppress Kv4.3 to determine the exact molecular composition of this residual ISA current.
Our studies also reveal several important aspects of ISA current regulation that will require further study to fully understand the underlying mechanisms. The first is that the amplitude of ISA is similar in hippocampal pyramidal neurons recorded in either dissociated culture or organotypic culture, at approximately 1000 pS/pF of current. It is therefore possible that, despite the significant differences in circuitry in these two culture systems, major aspects of the intrinsic genetic program that is regulating ISA are conserved. Future studies will be required to determine what factors are acting in these systems to regulate ISA current around this apparent set point.
A second interesting observation is that increases in ISA amplitude produced by over-expression of Kv4.2 are limited to around 170% of the normal current level regardless of vector system being used or whether the experiments were in dissociated or organotypic cultures. A further surprise is that this enhanced ISA current is indistinguishable from the normal ISA current. If Kv4.2 is expressed alone in the heterologous system, the biophysical properties of the expressed channels are quite distinct from native ISA (Jerng et al. 2005). Expressed alone, Kv4.2 inactivation occurs with three exponential kinetic components and recovery from inactivation requires hundreds of milliseconds. In addition, the kinetics of Kv4.2 inactivation become more rapid with depolarization, not slower as seen with native ISA. Only when Kv4.2 is co-expressed with KChIP and DPL auxiliary proteins do the functional properties of channels expressed in heterologous cells become similar to native ISA channels in kinetics and voltage-dependence (Jerng et al. 2004b, 2005). KChIP binding removes an N-type inactivation mechanism from the channel, the most of the channel inactivation to occur through a single closed-state mechanism that becomes slower with depolarization (Bahring et al. 2001; Gebauer et al. 2004; Zhou et al. 2004). DPL binding accelerates the kinetics of Kv4.2 channel gating to the rates similar to what is seen with native channels (Nadal et al. 2003; Jerng et al. 2005). Thus, we suggest that an intrinsic regulatory mechanism is acting in neurons to limit the surface expression of Kv4.2 subunits if they lack auxiliary subunit proteins.
Studies in heterologous systems show that Kv4.2 protein surface expression is strongly dependent on the presence of auxiliary subunits, with a 12–40-fold increase in current expression when Kv4.2 is co-expressed with KChIP and DPL auxiliary subunits (An et al. 2000; Bahring et al. 2001; Nadal et al. 2003; Jerng et al. 2004a; Zhou et al. 2004; Jerng et al. 2005). If similar effects are occurring in native neurons, then Kv4.2 channels that do not co-assemble with auxiliary subunits would only be put onto the cell surface with an efficiency 2.5% of those with auxiliary subunits. This low efficiency would effectively clamp the maximum level for over-expression of ISA to a point limited by the native availability of KChIP and DPL subunits in these neurons. Thus, a possible hypothesis to explain these results is that native KChIP and DPL expression levels are slightly higher than Kv4.2, sufficient for 70% more channels than are provided by native Kv4.2 subunit expression. An alternative possible explanation is that auxiliary proteins are not limiting at all, but native feedback mechanisms act to limit the amount of the protein increase onto the cell surface. Future studies will attempt to distinguish between these hypotheses by targeting auxiliary subunits with RNA interference, and over-expressing auxiliary subunits with or without Kv4.2 to determine a possible role for auxiliary protein expression in limiting the maximum surface density for ISA channels.
We would like to thank Dr David Johns for the Kv4 Dominant-negative mutant EGFP-tagged construct and Dr Daniel Johnston for his advice and support. This work was supported by NIH Grants NS37444 and MH48432 and NS31583.