Abstract
Modulation of voltage-gated sodium channels (VGSC) can have a major impact on cell excitability. Analysis of calmodulin (CaM) binding to GST-fusion proteins containing the C-terminal domains of Nav1.1-Nav1.9 indicates that some of the tetrodotoxin-sensitive VGSC isoforms, including NaV1.4 and NaV1.6, are able to bind CaM in a calcium-independent manner. Here we demonstrate that association with CaM is important for functional expression of NaV1.4 and NaV1.6 VGSCs. Disrupting the interaction between CaM and the C terminus of NaV1.4 and NaV1.6 channels reduced current amplitude by 99 and 62%, respectively. Overexpression of CaM increased the current generated by Nav1.4 and Nav1.6 C-terminal mutant constructs that exhibited intermediate current densities and intermediate binding affinities for CaM, demonstrating that this effect on current density was directly dependent on the ability of the C terminus to bind CaM. In addition to the effects on current density, calmodulin also was able to modulate the inactivation kinetics of Nav1.6, but not Nav1.4, currents in a calcium-dependent manner. Our data demonstrate that CaM can regulate the properties of VGSCs via calcium-dependent and calcium-independent mechanisms and suggest that modulation of neuronal sodium channels may play a role in calcium-dependent neuronal plasticity.
Introduction
Voltage-gated sodium channels (VGSCs) are the main carriers of current underlying the fast upstroke of action potentials in excitable cells. Modulation of VGSCs can regulate the function of the brain (Catterall, 2000) and muscle (Weiss and Horn, 1986). The activity of many channel proteins is regulated by calmodulin (CaM) (Levitan, 1999; Saimi and Kung, 2002), a 16.7 kDa protein that is expressed in virtually all eukaryotic cells. CaM can induce changes in target proteins via its binding per se and in response to changes in calcium concentration (Chin and Means, 2000). For example, the activity of sodium channels in Paramecium tetraurelia is CaM-dependent (Ling et al., 1992; Saimi and Ling, 1995), and CaM plays a crucial role in the modulation of calcium channels (Lee et al., 1999; Peterson et al., 1999; Zuhlke et al., 1999).
Several lines of evidence suggest that VGSCs might be modulated by CaM in muscle and neurons. A large number of CaM binding proteins have an isoleucine-glutamine (IQ) CaM binding motif (Bahler and Rhoads, 2002). Rhoads and Friedberg (1997) identified IQ motifs in the C termini of NaV1.1-1.3 and predicted that these VGSCs might bind CaM. Indeed, experiments that used a GST-fusion protein consisting of a 27-amino-acid fragment of the C terminus of NaV1.2 showed that CaM binds directly to the Nav1.2 IQ motif in a calcium-independent manner (Mori et al., 2000). Two recent studies suggest that CaM can affect the activity of NaV1.4 (Deschenes et al., 2002) and NaV1.5 (Tan et al., 2002) muscle VGSCs by binding to their IQ motifs.
IQ motifs have been identified in proteins that bind CaM permanently or transiently. Some proteins with IQ motifs bind CaM independently of calcium, whereas others selectively bind calcium/CaM or calcium-free CaM (Jurado et al., 1999; Bahler and Rhoads, 2002). The core of the IQ motif is often IQXXXRGXXXR, although the first residue is variable and can be ambiguous for isoleucine, leucine, or valine. The last residue reportedly can be either arginine or lysine, and the seventh residue is not well conserved. Thus a more generalized IQ motif might be [I,L,V]QXXXRXXXX[R,K] (Bahler and Rhoads, 2002). After aligning the C termini sequences of all known VGSC isoforms (see Fig. 1A), we observed that the key residues in the region of the C terminus sharing homology with a 23-residue-long IQ motif (Schultz et al., 2000) were conserved to varying degrees.
In this study we asked whether CaM binding to NaV1.6, a neuronal VGSC (Burgess et al., 1995; Schaller et al., 1995; Smith et al., 1998), had the same functional consequences as CaM binding to the skeletal muscle VGSC, NaV1.4 (Trimmer et al., 1989). Our novel data show that CaM regulates the current density of both NaV1.4 and NaV1.6. In addition, we show that CaM regulates the kinetic properties of NaV1.6 currents in a calcium-dependent manner. This work presents functional evidence for participation of VGSCs in the intricate CaM signaling network that is present within neurons (Marrion and Tavalin, 1998; Dolmetsch et al., 2001) and muscles (Hamilton et al., 2000).
Materials and Methods
Molecular biology. cDNAs corresponding to the C termini of all known voltage-gated sodium channels (Goldin et al., 2000) were generated by PCR. Rat brain or dorsal root ganglion (DRG) cDNA served as a template for NaV1.1, 1.2, 1.3, 1.4, 1.5, 1.8, and 1.9. NaV1.6 was amplified from mouse cDNA (Smith et al., 1998) and NaV1.7 from a human cDNA construct (Klugbauer et al., 1995). PCR products were subcloned into EcoRI/BamHI sites of the pGEX3X vector (Amersham Biosciences, Piscataway, NJ) to generate glutathione S-transferase (GST) fusion constructs.
All mutations were inserted into the respective sodium channel cDNA constructs by using the Quick Change XL (Stratagene, La Jolla, CA) mutagenesis kit, following the instructions of the manufacturer. Mutagenic primers were designed to introduce the appropriate base pair changes into the rat NaV1.4 and the mouse NaV1.6 sequences. Initially, both channels were made tetrodotoxin (TTX)-resistant by changing a pore tyrosine residue to a serine (Y401S in NaV1.4 and Y371S in NaV1.6) (Cummins et al., 2001) to allow currents to be isolated after expression in neurons and human embryonic kidney 293 (HEK293) cells. These two TTX-resistant channel constructs served as templates for all further mutagenesis. The plasmid Nav1.4-RBG4 (Ukomadu et al., 1992) was used for the Nav1.4 constructs. C-terminal truncation mutants of NaV1.4 were made by introducing a stop codon at amino acid position 1741 (DEL1) after the IQ motif and at position 1719 (DEL2) before the IQ motif. Primers were designed to change NaV1.4 amino acid residues 1727IQ 1728 to AQ, EQ, IA, IE, AA, and EE.
Construction of a mammalian expression vector encoding neuronal rat Nav1.6 channel. These studies used a cDNA construct that encodes the mouse NaV1.6 open reading frame. The complete open reading frame of mouse NaV1.6, generously provided by Dr. A. Goldin (Smith et al., 1998), was excised from a pLCT1 oocyte expression vector first by digestion with AatII and then by blunting the ends with T4 DNA polymerase, followed by digestion with XhoI to release the Nav1.6 insert. A mammalian expression vector pcDNA3.1 (Invitrogen, San Diego, CA), modified to render it a low-copy vector (Klugbauer et al., 1995), was digested with ApaI; the ends were blunted with T4 DNA polymerase and then digested with XhoI. In a subsequent ligation reaction the two DNA pieces were joined to form a 13 kb plasmid. Isolates that displayed the expected BamHI digestion pattern were verified by sequencing the entire NaV1.6 insert. A TTX-resistant phenotype of NaV1.6 (NaV1.6r) was produced by converting tyrosine 371 to serine (Y371S) as previously described for the Nav1.3 channel (Cummins et al., 2001). This point mutation was introduced into the channel sequence with the Quick Change XL mutagenesis kit (Stratagene), with two mutagenic primers designed according to the manufacturer's instructions. The mutant construct subsequently was verified by sequencing the whole Nav1.6 insert. A similar approach was used to mutate the NaV1.6 residues 1900LQ 1901 to EQ, LE, and EE.
A CaM-EGFP (enhanced green fluorescent protein) fusion construct was generated by PCR. Forward and reverse primers containing an EcoRI and BamHI restriction site, respectively, were used to amplify a wild-type CaM DNA fragment from a rat brain cDNA library, which then was subcloned into a previously EcoRI/BamHI-digested pEGFP-N1 vector (Clontech, Palo Alto, CA). A dominant-negative calcium-insensitive CaM mutant was generated by mutating the first aspartates of each of the four calcium-binding EF hand motifs to alanines (Putkey et al., 1989; Peterson et al., 1999). All constructs were verified by sequencing.
Expression and purification of GST-fusion proteins. For expression of GST-fusion proteins the appropriate plasmids (pGEX-C terminus-NaV 1.1-NaV1.9) were transformed into Escherichia coli DH5α (Stratagene). Fusion proteins were affinity purified on glutathione-Sepharose beads, as previously described (Liu et al., 1999).
GST pull-down assay. To examine the binding of CaM to the C termini of all of the sodium channels as well as the mutants of the NaV1.4 C terminus, in vitro, we incubated equal amounts of glutathione-Sepharose beads loaded with purified GST fusion proteins with 2 μg of bovine CaM in binding buffer AM-100 (Voit et al., 1997) at 4°C overnight. After being washed with binding buffer AM-100 four times, the bound proteins were released by boiling for 5 min in SDS-loading buffer and separated on a 12% SDS-PAGE. Subsequent immunoblot visualization was performed with monoclonal anti-CaM antibodies (Zymed, San Francisco, CA).
Coimmunoprecipitation. A HEK293 cell line that stably expressed rat Nav1.4 was grown in 100 mm dishes to 80% confluence. Sister cultures then were transfected with 30 μg of CaM-EGFP cDNA or EGFP cDNA, using 30 μl of Lipofectamine 2000 (Invitrogen) transfection agent. After 24 hr the cells in each dish were lysed in 500 μl of cell lysis buffer. Fractions of the supernatants were saved for loading on the gel, and the rest of the cell extracts was incubated with 5 μl of polyclonal anti-GFP antibody (Clontech). Protein A-agarose (Invitrogen) then was used to precipitate the protein complexes at 4°C overnight. After extensive washes the proteins were eluted in SDS loading buffer, and CaM was visualized by immunoblotting with a panspecific anti-sodium channel antibody (Upstate Biotechnology, Lake Placid, NY) used at a 1:1000 dilution.
Electrophysiology. Standard whole-cell patch-clamp techniques were used as previously described (Cummins et al., 1998). All cells were held at -120 mV for 5 min before the measurement of peak sodium current amplitude. The currents of the TTX-resistant constructs were recorded in the presence of 250 nm TTX to minimize contamination from endogenous HEK293 sodium currents. The pipette solution contained (in mm): 140 CsF, 1 EGTA, 10 NaCl, and 10 HEPES, pH 7.3. The bathing solution contained (in mm): 140 NaCl, 3 KCl, 1 MgCl2, 1 CaCl2, and 10 HEPES, pH 7.3. HEK293 cells were transfected by using the calcium-phosphate precipitation technique as previously described (Cummins et al., 1998).
Culture of DRG neurons. DRG neurons were cultured as described previously (Caffrey et al., 1992). Briefly, the L4 and L5 DRG ganglia were harvested from adult NaV1.8 null mice (Akopian et al., 1999). The DRG were treated with collagenase A (1 mg/ml) for 25 min and then collagenase D (1 mg/ml) and papain (30 U/ml) for 25 min, dissociated in DMEM and Ham's F-12 medium supplemented with 10% fetal bovine serum, and plated on glass coverslips. Nav1.8 null neurons were kept under standard tissue culture conditions for 5-7 d before biolistic transfections. We previously showed that Nav1.8 null DRG neurons do not express fast- or slow-inactivating TTX-resistant sodium currents (Cummins et al., 1999, 2001). Some Nav1.8 null DRG neurons do express persistent TTX-resistant sodium currents (Cummins et al., 1999), but these currents are typically <1 nA after several days in culture and run down quickly in whole-cell recording configuration; therefore, these persistent TTX-resistant currents are not significant under the culture and recording conditions used in the present study.
Biolistic transfection of Nav1.8 null DRG neurons. The Helios Gene Gun System (Bio-Rad Laboratories, Hercules, CA) was used for biolistic transfection of neurons. Nav1.6r or Nav1.4r DNA (10 μg) was mixed with 5 μg of GFP DNA in 50 μm spermidine and coprecipitated onto 1.6 μm gold particles with CaCl2. The DNA gold suspension was washed twice in 100% ethanol and resuspended in 0.05% polyvinylpyrrolidone in ethanol; it was used for coating the inner wall of 10 inches of Tefzel tubing (Bio-Rad Laboratories). The tubing was dried by using ultrapure nitrogen and was cut into ∼20 cartridges for the Helios Gene Gun. This process resulted in a density of 1 mg of gold particles per shot and 0.75 μg of total DNA per cartridge.
Immediately before biolistic transfection, the culture medium was removed from the Petri dish. The gene gun was held ∼2 cm above the cells, and a pressure of ∼120 psi was used to discharge the gold particles. A 70 μm nylon mesh (Small PartsMiami Lakes, FL) was placed just in front of the Helios Gene Gun barrel liner to achieve a more uniform distribution of gold particles (Wellmann et al., 1999). Within 24 hr the cells usually showed expression of GFP, indicating a successful biolistic transfection. Electrophysiological studies were conducted 18-48 hr after transfection, and most of the cells that expressed GFP also expressed fast-inactivating TTX-resistant sodium currents. Because these currents are not observed in untransfected Nav1.8 null neurons or Nav1.8 null neurons transfected with GFP alone, this confirmed that most of the cells that expressed GFP also had been cotransfected successfully with the Nav1.6 TTX-resistant channel.
Statistical analysis. One-way ANOVAs were used to test for differences between the experimental groups. The difference between pairs of individual data sets was tested by Fisher's least significant difference (LSD) multiple comparison test, and p values < 0.05 were considered to be significant. Where appropriate, data were compared with Student's t test. Data are expressed as the mean ± SEM.
Results
Several members of the VGSC family are able to bind CaM directly
Many proteins, including voltage- and ligand-gated channels, are regulated by CaM (Levitan, 1999; Zhu et al., 2001). Some proteins bind to CaM in a constitutive manner, whereas others are able to do so if calcium ions are present in a sufficiently high concentration. To determine to which category the VGSCs belong, we examined CaM binding in the presence and the absence of Ca2+ions. In analogy to the experiments conducted with calcium channels (Peterson et al., 1999), we generated cDNAs of GST-fusion proteins of the complete C terminus of all of the known members of the VGSC family (Goldin et al., 2000), NaV1.1 through NaV1.9. We used the whole C terminus rather than shorter peptide fragments because studies have shown that residues outside the immediate CaM binding region can have significant effects on binding affinity (Peersen et al., 1997). After expression and purification we incubated the GST-fusion proteins overnight with 2 μg of bovine CaM in the presence and absence of calcium ions. Because binding occurred under these in vitro conditions, we could exclude the possibility that any other factors might have mediated the observed interaction. Immunoblotting with an anti-CaM antibody revealed isoform-specific differences in CaM binding (Fig. 1B). Under both high and low calcium conditions the fusion proteins of NaV1.2, NaV1.4, and NaV1.6 exhibited very strong binding to calmodulin, suggesting that, like Nav1.2 (Mori et al., 2000), NaV1.4 and NaV1.6 also might be considered as high-affinity targets for CaM. The C termini of NaV1.1 and NaV1.3 showed weaker binding but also were able to bind CaM in the presence and absence of Ca2+ ions. The C terminus of Nav1.7 exhibited weak binding to CaM only in the low calcium condition. The three TTX-resistant sodium channels (NaV1.5, NaV1.8, and NaV1.9) did not show detectable binding of CaM to their C termini in our assay. This assay indicates that, whereas the TTX-sensitive VGSC isoforms (with the noted exception of Nav1.7) bind CaM to their C termini in the presence and absence of Ca2+ ions, the TTX-resistant isoforms probably exhibit a much lower affinity for CaM.
CaM binds to NaV1.4 in vivo
CaM binding to VGSCs has not been shown to occur in mammalian cells yet, raising the possibility that CaM may not bind to full-length channels. To address this issue, we performed a coimmunoprecipitation with the skeletal muscle sodium channel NaV1.4, which showed high-affinity binding to CaM in the GST pull-down assay and has been well characterized (Ukomadu et al., 1992). We transfected a HEK293 cell line that stably expressed the rat NaV1.4 channel with a plasmid encoding a CaM-EGFP fusion protein. It has been shown previously that EGFP does not interfere with normal CaM-target interactions (Erickson et al., 2001). A pull-down assay for the sodium channel protein complex, using a polyclonal anti-GFP antibody and a panspecific anti-sodium channel antibody for visualization, showed that CaM-EGFP was part of the sodium channel complex (Fig. 1C). This demonstrates that CaM truly can bind to the full-length NaV1.4 protein in an in vivo environment. Electrophysiological examination of the currents produced by NaV1.4 in this experiment revealed no change of any of its properties in comparison to the sister cultures expressing NaV1.4 alone (data not shown). However, this cannot be interpreted as evidence that CaM does not alter the functional properties of Nav1.4 channels. The data in Figure 1B suggest that Nav1.4, like Nav1.2, binds CaM with a high affinity. Because CaM is thought to be expressed in all eukaryotic cells, it is possible that the endogenous levels of CaM in HEK293 cells are sufficient to saturate Nav1.4 channels; therefore, the expression of additional CaM has no further effect.
CaM directly modulates NaV1.4 function
To investigate further the functional implications of the CaM-VGSC complex, we constructed two different C-terminal deletion mutants (Fig. 2A) and compared their sodium currents with those produced by the full-length NaV1.4. In construct DEL1 we deleted the amino acids from the end of the C-terminal up to, but not including, the IQ motif. In construct DEL2 we deleted an additional 23 amino acids in the C terminus that comprise the IQ motif. Channel constructs were transiently expressed in HEK293 cells to determine whether the C-terminal region containing the IQ motif was functionally relevant. All of our electrophysiological measurements were conducted by using NaV1.4 channels mutated to be TTX-resistant by changing the tyrosine at position 401 to serine (Y401S) and by recording in the presence of TTX to help differentiate our recombinant channel currents from the endogenous TTX-sensitive currents that are often present at low, but detectable, levels in HEK293 cells (Cummins et al., 1993). In this and subsequent experiments the term “wild-type Nav1.4” will be used to refer to the TTX-resistant variant Nav1.4-Y401S.
The currents produced by construct DEL1 (Fig. 2B) had activation and inactivation kinetics and peak amplitudes (n = 15) that were identical to those of the wild-type Nav1.4 channel (n = 44), indicating that the distal portion of the C terminus is not vital for normal channel function in HEK293 cells. In contrast, the DEL2 construct produced little or no sodium current (n = 18) when tested with the same voltage-clamp protocol as the wild-type channel and construct DEL1, showing that the region containing the IQ motif is critical for the generation of inward current by NaV1.4 (Fig. 2B). To determine whether the DEL2 construct produced protein levels comparable to wild-type NaV1.4 and DEL1, we performed a Western blot with protein extracted from cells that were transfected, with the same amounts of DNA for all three constructs (Fig. 2C). Similar amounts of protein were produced by all three constructs, demonstrating that, although the IQ motif is necessary for NaV1.4 current, it is not required for protein expression.
Prompted by this drastic effect of the DEL2 deletion on NaV1.4 current density, we wanted to refine further our analysis of the importance of the IQ motif. Mutations that change the hydrophobicity and charge of the core residues isoleucine and glutamine of the CaM binding motif have been used previously to study the interaction of CaM with its targets (Peterson et al., 1999; Zuhlke et al., 2000; Deschenes et al., 2002). We designed Nav1.4 constructs in which the corresponding amino acids were mutated to either alanines or glutamic acids and tested their electrophysiological properties. The isoleucine and the glutamine of the IQ motif (Fig. 1A) first were changed individually to either an alanine or a glutamic acid; then both residues were mutated together (Fig. 3A). Changing the isoleucine at position 1727 to the similar alanine (referred to as AQ) induced no significant changes in current properties when compared with the wild-type (IQ) channel. However, introducing a charged glutamic acid (EQ) at position 1727 reduced the peak sodium current by >90% (Fig. 3D). The remaining current is similar in amplitude to the endogenous TTX-resistant sodium current often observed in untransfected HEK293 cells (Cummins et al., 1999). In contrast, altering the second residue, glutamine 1728, to either alanine (IA) or glutamic acid (IE) reduced sodium currents by ∼50% (Fig. 3B). Interestingly, neither the voltage dependence of activation and steady-state inactivation nor any of the time constants that were measured were different for the IE or for any other mutant that we compared with wild-type NaV1.4 (IQ) (Fig. 3C). The double mutations, I1727A/Q1728A (AA) and I1727E/Q1728E (EE), reduced sodium currents to background levels (Fig. 3B,D), similar to the current reduction observed with construct DEL2 (Fig. 2B). This shows that the current reduction is not attributable to the addition of negative charges into the C terminus alone and suggests that it is attributable to disruption of the IQ motif.
To determine whether the observed amplitude changes were correlated with altered binding of CaM to the channel, we introduced the glutamic acid mutations EQ, IE, and EE into the GST-fusion protein of the NaV1.4 C terminus and tested their ability to bind CaM (Fig. 3E). Changing IQ to IE reduced binding, and changing IQ to EQ or EE seemed to prevent binding of CaM to the C terminus completely. This correspondence between CaM binding of the different IQ mutants and peak current amplitudes provides evidence for a direct link between CaM binding and the regulation of current amplitude and suggests that the current reduction was not attributable to CaM-independent interactions.
If the hypothesis that the reduced current levels observed with the IE and IA mutant channels were the result of a lower affinity of CaM for the C terminus of these channels is correct, then increasing the intracellular CaM concentration should increase the current produced by the IE and IA mutant Nav1.4 channels. In contrast, increasing CaM concentration would not be predicted to increase the current produced by the EE mutant, because this double mutation appeared to eliminate CaM binding to the C terminus of Nav1.4. Therefore, we tested whether over-expression of CaM was able to rescue the loss of current observed with the IQ mutants. We coexpressed the CaM-EGFP construct with wild-type NaV1.4, the IE and IA mutant constructs that produced intermediate current amplitudes, and the EE construct, which produced little or no current (Fig. 3B). Higher intracellular CaM concentrations had no significant influence on current amplitude (Fig. 3D) or voltage dependence of the wild-type channel (data not shown). Although we did not observe an increase in current with the EE mutant, which eliminated CaM binding in the GST-fusion protein assay, overexpression of CaM boosted the current of the intermediate IE and IA constructs to wild-type current levels (Fig. 3B,D). These data confirm that CaM can regulate NaV1.4 sodium current density directly by binding to the C terminus.
CaM also modulates NaV1.6 current amplitude
We next asked whether neuronal VGSC isoforms might be regulated by CaM in the same manner as Nav1.4. The initial GST pull-down assay indicated that NaV1.6, an isoform that is very abundant in brain (Burgess et al., 1995; Tzoumaka et al., 2000), binds CaM with an affinity comparable to that of NaV1.4 and Nav1.2. The IQ motif of NaV1.6 is slightly different from that of the other VGSCs (Fig. 1A) in that NaV1.6 has a leucine instead of an isoleucine at the position corresponding to I1727 of NaV1.4. However, this residue of the IQ motif has been shown to be variable in conventional myosins, and isoleucine, leucine, or valine generally is considered equivalent at this position (Bahler and Rhoads, 2002). In addition, the I1727L mutation had no effect on the functional properties of Nav1.4 (data not shown). To test the functional consequences of CaM binding to the IQ motif of NaV1.6, we mutated two of the core residues of the IQ motif in NaV1.6, residues L1900 and Q1901, to glutamates. We compared the properties of wild-type NaV1.6 channels (LQ) with NaV1.6-Q1901E (LE) and NaV1.6-L1900E-Q1901E (EE) mutant channels. Because NaV1.6 channels express poorly in HEK cells (0.31 ± 0.05 nA; n = 30), we used a TTX-resistant variant of NaV1.6 (NaV1.6-Y371S) expressed in DRG neurons from NaV1.8 null mice (Akopian et al., 1999). After 5 d in culture these neurons lack TTX-resistant currents and therefore provide a very suitable expression system for expression of sodium channels that have been mutated to be TTX-resistant (Cummins et al., 2001).
At 24 hr after biolistic transfection large (peak amplitude, 40.8 ± 4.0 nA; n = 31) NaV1.6 sodium currents (Fig. 4A) were recorded in the presence of 250 nm TTX. The currents produced by the LE construct were 74% smaller than wild-type (LQ) currents (Fig. 4B). In contrast to the effect of the IQ mutations on NaV1.4, the Nav1.6 construct with the EE double mutation reduced NaV1.6 current amplitudes down to only ∼37% of control (Fig. 4A,B). ANOVA plus pairwise comparisons (Fisher's LSD multiple comparison test) revealed that the mutant current amplitudes were significantly smaller than wild-type current amplitude but that the difference between the current amplitudes produced by LE and EE mutants was not significant. To determine whether the difference in the effect of the EE double mutation on NaV1.4 and NaV1.6 current amplitude was the result of the different expression backgrounds (HEK293 cells vs DRG neurons), we expressed NaV1.4 and NaV1.4-EE channels in NaV1.8 null DRG neurons. Because we were using TTX-resistant constructs, we again were able to isolate the transfected currents by using 250 nm TTX. Although large TTX-resistant sodium currents were recorded from neurons transfected with NaV1.4 channels, virtually no current was observed in neurons transfected with NaV1.4-EE channels (Fig. 4B). It is not clear what accounts for the difference in the effect of the EE double mutation on NaV1.4 and NaV1.6 current amplitude, but it could reflect additional binding sites for CaM on NaV1.6 channels, as has been reported for some L-type calcium channels (Pitt et al., 2001).
In analogy to the study of NaV1.4, we wanted to determine whether the modulation of Nav1.6 current amplitude by the IQ motif mutations depended on the reduced binding of CaM to the channel. Therefore, we tested whether overexpression of CaM could rescue the reduced current of the LE construct. The LE current amplitude was increased by 118% (Student's t test, p < 0.001) when more CaM was available to the channel. As with NaV1.4, this change in current density was not associated with changes in the voltage dependence of activation or inactivation.
Time course of inactivation varies between wild-type NaV1.6 (LQ) and EE mutant currents
Although the activation time constant τm did not differ between the LQ and EE NaV1.6 channel constructs, there was a significant difference (Student's t test, p < 0.05) in the inactivation time constant τh between LQ and the EE constructs (Fig. 5). The EE double mutation slowed NaV1.6 inactivation by 48% at -30 mV and 51% at +20 mV. The rate of fast inactivation was slower for the EE mutant channels at all voltages from -30 to +40 mV. Because shifts in the voltage dependence of channel gating can contribute to changes in inactivation kinetics, we compared the voltage dependence of activation and steady-state inactivation for NaV1.6 LQ and EE constructs. The midpoints for activation (determined with a Boltzmann function) were not significantly different for LQ (-25.4 ± 1.0 mV; n = 7) and EE (-22.9 ± 2.4 mV; n = 8) Nav1.6 currents. The midpoints for steady-state inactivation were also similar for LQ (-64.3 ± 2.1 mV; n = 10) and EE (-65.5 ± 2.0 mV; n = 9) Nav1.6 currents. Thus shifts in the voltage dependence of channel gating do not seem to contribute to the differences in inactivation kinetics.
Calcium/calmodulin switches NaV1.6 from a fast into a slow mode
In the GST pull-down assay we showed that CaM binds to NaV1.6 in the presence and absence of calcium ions. However, CaM is considered the prototype of a family of intracellular calcium sensor proteins that confer calcium sensitivity to their targets (Chin and Means, 2000; Burgoyne and Weiss, 2001). Because reducing the binding of CaM to NaV1.6 changes its inactivation kinetics, we asked whether this reflects a physiologically relevant CaM-dependent mechanism that also could be regulated by changes in intracellular calcium concentration. All of the experiments described above were conducted with low intracellular calcium in which the pipette solution contained 1.1 mm EGTA and no added calcium. Similar results were obtained with 20 mm BAPTA and no added calcium in the pipette. To examine NaV1.6 current properties at a high [Ca2+]i, we added 1.1 mm calcium to the pipette solution that contained 1.1 mm EGTA, which raised the free calcium level to ∼10 μm. We also examined the effect of a CaM mutant in which the aspartates in the first position of the four calcium-binding EF hand motifs had been replaced by an alanine to render it calcium-insensitive. This mutant usually is referred to as a dominant-negative form of CaM and is abbreviated as CaM1234 (Peterson et al., 1999; Pitt et al., 2001).
We first examined the effect of coexpression of wild-type CaM with NaV1.6 on NaV1.6 current characteristics. Coexpression of wild-type CaM with NaV1.6 did not alter the properties of currents recorded with low [Ca2+]i. Raising [Ca2+]i in the presence of overexpressed CaM did not alter the voltage dependence of activation (Fig. 6A) or steady-state inactivation (Fig. 6B). However, with high [Ca2+]i the currents recorded from cells transfected with NaV1.6 and CaM exhibited slower time constants of inactivation (Fig. 6C,D). On the other hand, when NaV1.6 channels were coexpressed with CaM1234, the inactivation kinetics were fast in the presence of both high and low [Ca2+]i (Fig. 6D). Thus high [Ca2+]i with wild-type CaM, but not with the calcium-insensitive CaM1234, had the same effect on the rate of inactivation as did the EE double mutation. This indicates that calcium-free CaM can enhance the rate of Nav1.6 inactivation.
For comparison, we also tested the calcium sensitivity of NaV1.4 channels. Increasing [Ca2+]i had no effect on the voltage dependence of activation (Fig. 7A) or steady-state inactivation (Fig. 7B) of Nav1.4 channels expressed in HEK293 cells. The inactivation kinetics of NaV1.4 currents expressed in either HEK293 cells (Fig. 7C) or NaV1.8 null DRG neurons (data not shown) also were not altered by changes in [Ca2+]i. This indicates that the calcium-dependent calmodulin regulation of VGSC inactivation kinetics is isoform-specific.
Recently, Deschenes et al. (2002) reported that calcium/CaM shifted steady-state inactivation of Nav1.4 currents in the negative direction by ∼5 mV. Because we did not observe any shift in steady-state inactivation under our conditions, we performed additional sets of experiments that used conditions similar to those used by Deschenes and colleagues. The first set of experiments used the intracellular and bath solutions (Fig. 8) described by Deschenes et al. (2002) with no calcium buffering capacity in the pipette solution. No difference was observed in the voltage dependence of activation (Fig. 8A) or steady-state inactivation (Fig. 8B) between cells transfected with either Nav1.4 plus CaM or Nav1.4 plus EGFP. In the second set of experiments that used these solutions, the intracellular free calcium was buffered to ∼10 μm with the addition of 1.1 mm EGTA and 1.1 mm CaCl2 to the pipette solution. In this set of experiments we recorded from HEK293 cells that stably expressed Nav1.4 channels with or without the addition of 10 μm CaM (Calbiochem) to the pipette solution. At 10 min after establishing the whole-cell recording configuration, we measured the voltage dependence of activation (Fig. 8C) and steady-state inactivation (Fig. 8D). Calcium/calmodulin did not alter the voltage dependence of Nav1.4 currents under these conditions. Therefore, we conclude that calcium/calmodulin does not alter the voltage-dependent or kinetic properties of Nav1.4 currents.
Discussion
In this study we describe the modulation of two VGSC isoforms, NaV1.4 and NaV1.6, by CaM and intracellular calcium. We show that functional expression of Nav1.4 current is critically dependent on the ability of Nav1.4 to bind CaM to its C terminus, indicating that this is a novel mechanism for the regulation of sodium current amplitude. Mutations in the IQ motif of NaV1.6 channels also reduced NaV1.6 current amplitude, although to a smaller extent than NaV1.4 channels. We did not find any evidence that CaM modulated the voltage dependence of activation or inactivation of either Nav1.4 or Nav1.6 channels. However, we found that changes in the intracellular calcium ion concentration altered the inactivation kinetics of NaV1.6 currents via a CaM-dependent mechanism. These data show that CaM can regulate the properties of VGSCs in isoform-specific ways and via both calcium-dependent and calcium-independent mechanisms.
The regulation of Nav1.4 and Nav1.6 current density by CaM was surprising, but not unprecedented. CaM appears to be an important regulator of several different ionic currents; recently, Saimi and Kung (2002) proposed that CaM could be considered a channel subunit in itself. The high voltage-activated sodium channels of Paramecium are critically dependent on CaM for activation (Saimi and Ling, 1995). CaM has been shown to interact with the C terminus of SK4 potassium channels, and this interaction controls channel assembly and surface expression of SK4 channels (Joiner et al., 2001). Functional expression of KCNQ2/Q3 potassium channels also requires CaM binding, although in this case CaM is not required for targeting of KCNQ2 to the cell membrane despite the fact that KCNQ2 seems to bind CaM constitutively in a calcium-independent manner (Wen and Levitan, 2002).
Although CaM is expressed at relatively high concentrations (1-100 μm) in many cell types, CaM binding proteins may outnumber CaM by as much as 2:1. Persechini and Cronk (1999) estimated that, although under normal conditions high-affinity (KD ≤ 10 nm) CaM binding proteins are able to bind CaM, low-affinity (KD ≥ 100 nm) CaM binding proteins are able to bind CaM only if the local concentration of free CaM can be increased significantly. Mori et al. (2000) reported that affinity of the IQ motif of Nav1.2 for CaM was 5 nm; therefore, Nav1.2 would be considered a high-affinity CaM binding protein. Our data suggest that Nav1.4 and Nav1.6 also might be high-affinity CaM binding proteins (Fig. 1B) and might bind CaM constitutively. Indeed, although eliminating CaM binding to the IQ motifofNav1.4orNav1.6 channels had major effects on current density, the overexpression of CaM did not seem to alter the properties of wild-type Nav1.4 or Nav1.6 currents, suggesting that the channels might be saturated with CaM. The fact that overexpression of CaM was able to increase the current amplitude in cells expressing Nav1.4 or Nav1.6 channels with IQ mutations that reduced, but that did not eliminate, current and CaM binding provides compelling evidence for a direct regulation of sodium current amplitude by CaM. Interestingly, a recent study indicated that a mutation identified in the C terminus of the human Nav1.2 channel (R1902C) from an individual with autism might reduce the binding affinity of Nav1.2 for calcium/calmodulin (Weiss et al., 2003).
Our GST-fusion protein assays indicated that the C termini of Nav1.1, Nav1.3, and Nav1.7 bound CaM with a lower affinity than the other TTX-sensitive isoforms. Although the IQ motifs of the various sodium channel isoforms are fairly well conserved, there are differences in this region (for example, Nav1.3 has an asparagine at a position at which the other isoforms have an alanine; Fig. 1A), and these differences could alter CaM binding. However, it is likely that other regions in the C termini of the different isoforms also influence CaM binding. It will be interesting to determine whether CaM also can regulate the current density produced by the TTX-sensitive VGSC isoforms other than Nav1.4 and Nav1.6. None of the three known TTX-resistant isoforms (Nav1.5, Nav1.8, and Nav1.9) showed detectable CaM binding in our assay. Two previous studies have suggested that Nav1.5 might bind CaM. Tan et al. (2002) reported that a 20-amino-acid peptide containing the IQ motif of Nav1.5 was able to bind CaM, and Deschenes et al. (2002) reported that a truncated form of the C terminus of Nav1.5 that contained 53 amino acids was able to bind CaM. Our experiments were conducted by using the complete C terminus of Nav1.1-Nav1.9 for the in vitro binding studies, and, because regions outside the IQ motif can influence CaM binding (Peersen et al., 1997), conformational changes in the full-length C terminus might account for the lack of CaM binding to Nav1.5 in our assay. Interestingly, Tan et al. (2002) reported that the 20 AA peptide containing the Nav1.5 IQ motif showed a lower affinity for CaM than the 290-309 CaM antagonist peptide derived from CaM kinase II. Because the 290-309 peptide has an affinity for CaM of 52 nm (Payne et al., 1988), this indicates that the affinity of the Nav1.5 IQ motif for CaM is >10-fold lower than that of the Nav1.2 IQ motif. Based on these data, Nav1.5 would be considered a low-affinity binding protein and would not be expected to bind CaM constitutively. Although the study by Tan et al. (2002) suggested that calcium/CaM might decrease the amplitude of Nav1.5 currents by ∼5-15% by regulating an intermediate form of slow inactivation, Deschenes et al. (2002) concluded that Nav1.5 currents are modulated only indirectly by CaM via a CaM-dependent kinase.
We did not see any effect of CaM overexpression on the voltage-dependent properties of Nav1.4 or Nav1.6 currents. In contrast, Deschenes et al. (2002) reported that overexpression of CaM in HEK293 cells produced a -6 mV shift in the voltage dependence of steady-state inactivation of Nav1.4 currents (Deschenes et al., 2002). Whereas we used 1.1 mm EGTA to buffer intracellular free calcium in our control experiments, Deschenes et al. (2002) did not include any calcium buffers in their intracellular solution for their control conditions. However, we also did not observe any effect of overexpression of CaM on Nav1.4 currents when we used the solutions described by Deschenes and colleagues without intracellular calcium buffering. In unbuffered solutions the free calcium levels can be quite variable and even can be ≥10 μm (Pallotta et al., 1992). Therefore, we also examined the effect of adding CaM to the Deschenes et al. (2002) pipette solution when free calcium was buffered to ∼10 μm with EGTA. Under these conditions the additional CaM had no effect on the voltage dependence of Nav1.4 currents. We did not find any evidence that the properties of Nav1.4 were regulated by CaM in a calcium-dependent manner in any of the experiments that we conducted.
Our data showed that the functional properties of Nav1.6 currents are regulated by calcium/CaM, which slowed inactivation by ∼50%. However, because the Nav1.6-EE currents also exhibited slowed inactivation, our data indicate that calcium-free CaM enhances the rate of inactivation, and, when calcium binds to CaM, calcium/CaM disrupts this enhancement, thereby effectively slowing inactivation. By disrupting the binding of CaM to the IQ motif, the EE mutation effectively would prevent the calcium-free CaM enhancement of inactivation. Although in this proposed mechanism calcium-free CaM is the active molecule, which is the case for other actions of CaM (Jurado et al., 1999), this still represents a calcium-dependent modulation of sodium current properties. This is the first description of a calcium-dependent CaM modulation of a neuronal sodium channel isoform. CaM has been shown to confer calcium sensitivity to several other transmembrane channel proteins (Saimi and Kung, 1994; Levitan, 1999). The fact that we are now able to observe a similar regulatory mechanism for Nav1.6, a neuronal VGSC that is present along axons (Caldwell et al., 2000) and extends to axon terminals (Black et al., 2002), has potentially interesting implications for the control of neuronal excitability and synaptic transmission by calcium/CaM. The calcium/CaM-dependent slowing of Nav1.6 inactivation kinetics could prolong action potential duration, and in nerve terminals this type of modulation might enhance neurotransmitter release.
Conclusions
Our data show that CaM binding to the C terminus of VGSCs can be critical for obtaining functional sodium current expression and may provide a mechanism for regulating the amplitude of sodium currents in a concentration-dependent manner. Although our experiments with GST-fusion proteins indicate that CaM binding to the C terminus of VGSCs is not dependent on calcium concentration, calcium and CaM are known to play important roles in excitation contraction coupling in muscle (Hamilton et al., 2000) and in controlling the excitability of neurons (Sola et al., 2001), and the interaction of the CaM-VGSC complex with other proteins in neurons and muscle cells may be modulated by calcium. In addition, although we did not observe any calcium/CaM-dependent regulation of Nav1.4 currents, the inactivation kinetics of Nav1.6 currents were modulated by calcium/CaM. The array of mechanisms that involve finely tuned Ca2+ concentrations just below the cell membrane is growing, and our data add a new mechanism, mediated by CaM, to the ways in which VGSCs can be regulated.
Footnotes
This work was supported in part by grants from the National Multiple Sclerosis Society and the Rehabilitation Research and Development Service and Medical Research Services, Department of Veterans Affairs, and by gifts from the Paralyzed Veterans of America and Eastern Paralyzed Veterans Association (EPVA). R.H. was an EPVA Spinal Cord Research Fellow. We thank Bart Tofness for providing us with excellent technical assistance.
Correspondence should be addressed to Dr. Theodore R. Cummins, Stark Neuroscience Research Institute, Indiana School of Medicine, 950 West Walnut, Indianapolis, IN 46202. E-mail: trcummin{at}iupui.edu.
C. Liu's present address: Department of Orthopaedic Surgery, New York University School of Medicine, New York, NY 10003.
Copyright © 2003 Society for Neuroscience 0270-6474/03/238261-10$15.00/0