Modulation of the cardiac sodium channel Nav1.5 by fibroblast growth factor homologous factor 1B.

We have previously shown that fibroblast growth factor homologous factor 1B (FHF1B), a cytosolic member of the fibroblast growth factor family, associates with the sensory neuron-specific channel Na(v)1.9 but not with the other sodium channels present in adult rat dorsal root ganglia neurons. We show in this study that FHF1B binds to the C terminus of the cardiac voltage-gated sodium channel Na(v)1.5 and modulates the properties of the channel. The N-terminal 41 amino acid residues of FHF1B are essential for binding to Na(v)1.5, and the conserved acidic rich domain (amino acids 1773-1832) in the C terminus of Na(v)1.5 is sufficient for association with this factor. Binding of the growth factor to recombinant wild type human Na(v)1.5 in human embryonic kidney 293 cells produces a significant hyperpolarizing shift in the voltage dependence of channel inactivation. An aspartic acid to glycine substitution at position 1790 of the channel, which underlies one of the LQT-3 phenotypes of cardiac arrythmias, abolishes the interaction of the Na(v)1.5 channel with FHF1B. This is the first report showing that interaction with a growth factor can modulate properties of a voltage-gated sodium channel.

Voltage-gated Na ϩ channels, which are essential for the generation of action potentials and cell excitability, are complexes of a pore-forming ␣-subunit and auxiliary ␤-subunits (1). Ten distinct pore-forming ␣-subunits of sodium channels have been identified in vertebrates (2). Dysfunction or dysregulation of sodium channels can cause a number of disorders, including life-threatening cardiac arrhythmias (3,4) and neuropathic pain (5). Thus, factors that modulate sodium channel function or subcellular localization are of interest from both a pathophysiological and a therapeutic standpoint.
Different classes of proteins have been shown to interact with several sodium channels and influence their surface expression and/or the properties of their Na ϩ currents. Auxiliary ␤-subunits affect the assembly, trafficking, and electrophysiological activities of ␣-subunits (6) and have recently been shown to link sodium channels to extracellular matrix proteins via their association with neurofascin (7). The neuronal adhesion molecule contactin/F3 increases the membrane insertion of sodium channels by direct (8) or indirect (9) association with the ␣-subunit of the channels. The association with contactin/F3 may also mediate the interaction of the channel with the extracellular protein tenascin (8). Annexin II light chain (p11) binds to the N terminus of sensory neuron-specific sodium channel Na v 1.8 and enhances its functional expression (10). The C terminus of the skeletal muscle and cardiac muscle ␣-subunits, Na v 1.4 and Na v 1.5, respectively, contain a PDZbinding motif that links these channels to syntrophin (11).
Channel partners can modulate the Na ϩ current either as a direct effect of binding or by inducing post-translational modifications, for example, phosphorylation. Peak Na ϩ current amplitude in cultured hippocampal neurons is reduced upon activation of dopamine (D1) receptors via cAMP-dependent protein kinase that is anchored to the ␣-subunit by the channel partner AKAP15 (12,13). Receptor protein-tyrosine phosphatase ␤ interacts with brain sodium channels, phosphorylates tyrosine residue(s) of the channels, and causes a depolarizing shift in the voltage dependence of inactivation of recombinant Na v 1.2 (14). The Ca 2ϩ -binding protein calmodulin interacts with the C terminus of Na v 1.2 (15), Na v 1.5, Na v 1.4 (16,17), and Na v 1.6 1 and modulates the Na ϩ currents in an isoform-specific manner.
Fibroblast growth factor homologous factor 1B (FHF1B, also known as FGF12B) 2 belongs to the FGF family of growth factors (18). FHF1-4 (FGF11-14) form a subfamily whose members lack canonical signal peptides and are not secreted into the medium of transfected cells (18 -21). Recently, FHF1 and FHF2 have been shown to bind to the scaffold protein Islet of Brain 2 (IB2) and recruit the mitogen-activated protein kinase p38␦ to the signaling complex (22). Previously, we reported the interaction of FHF1B with a sensory neuron-specific, tetrodotoxin-resistant (TTX-R) voltage-gated sodium channel Na v 1.9 (23). The functional significance of this interaction has yet to be determined. FHF1B did not interact with the C termini of the sodium channels Na v 1.7 and Na v 1.8, which, similar to Na v 1.9, are expressed in the sensory neurons of the dorsal root ganglia (23). In this study, we demonstrate that FHF1B associates with another TTX-R sodium channel, the cardiac channel Na v 1.5.
We show that FHF1B binds to Na v 1.5 both in vitro and in vivo and that the co-expression of FHF1B with wild type human Na v 1.5 causes a hyperpolarizing shift in steady-state inactivation of this channel.
The bacterial expression vector pGEX-3X (Invitrogen) was used to produce recombinant proteins in Escherichia coli. A fragment encoding the C terminus of Na v 1.5 was subcloned in-frame into the BamHI/EcoRI sites of pGEX-3X to produce the plasmids pGEX-Na v 1.5C.
A construct of full-length human Na v 1.5 was a generous gift from Dr. Mike O'Leary. The complete open reading frame and flanking untranslated sequences were cloned into the HindIII/XbaI sites of the vector pcDNAI. The mutant D1790G was generated using site-directed mutagenesis. The forward mutagenic primer (5Ј-AGCCCCTGAGTGAGGA-CGGCTTCGATATGTTCTATG-3Ј) and the reverse mutagenic primer (reverse complement sequence of the forward primer) were used in the QuikChange mutagenesis system (Stratagene); the single nucleotide substitution G, underlined and in bold type, causes the D1790G LQT-3 mutation. The sequence of the entire insert was verified to eliminate the possibility of additional unintended substitutions.
All of the constructs were verified by sequencing their inserts. Sequencing was carried out at the Howard Hughes Medical Institute/Keck Biotechnology Resource Laboratory at Yale University. Sequence analysis was done using Lasergene (DNASTAR) and BLAST (National Library of Medicine) software.
Yeast Two-hybrid Assay for Protein-Protein Interactions-Three independent colonies were analyzed for the interaction in yeast between two proteins, of which one was fused to the Gal4 DNA binding domain and the other was fused to the VP16 transactivation domain. The procedures of Vojtek et al. (24) and Hollenberg et al. (25) were followed for (i) growing and transforming the yeast strain MAV203 with the selected plasmids and (ii) testing by the ␤-galactosidase assay and by growth phenotypes on (SD leu/trp/his/ura/3AT ϩ ) plates and on plates containing 5-fluoroorotic acid (SD leu/trp/5FOA ϩ ).
In Vitro Binding Assay-For expression of GST fusion proteins, the plasmid pGEX-Na v 1.5C was transformed into E. coli DH5 (Invitrogen). The fusion proteins were affinity-purified on glutathione-agarose beads as previously described (26). To examine the binding of FHF1B and its derivatives to the C terminus of rNa v 1.5 in vitro, glutathione-Sepharose beads (50 l) preincubated with purified GST (0.5 g), serving as control, or GST-Na v 1.5C (0.5 g) were incubated with extract (500 g of protein) of HEK293 cells transfected with an expression plasmid encoding either full-length FHF1B-GFP or its derivatives (i.e. FHF1B⌬N4-GFP, FHF1B⌬N41-GFP, FHF1B⌬N76-GFP, and FHF1B-⌬N142-GFP), as indicated, in 150 l of buffer AM (10 mM Tris, pH 7.9, 10% glycerol, 1 mM MgCl 2 ) supplemented with 100 mM KCl and 0.5 mg/ml bovine serum albumin (26). The bound proteins were denatured in sample buffer and separated by 12% SDS-PAGE, and GFP fusion proteins were detected by immunoblotting with anti-GFP antibodies (Clontech).
Co-immunoprecipitation-HEK293 cells were grown in 100-mm dishes in Dulbecco's modified Eagle's medium containing heat-inactivated 10% fetal bovine serum. After reaching 50% confluence, the cultures were co-transfected with the mammalian expression plasmids pCMV-Na v 1.5 and pEGFP-N1 or pFHF1B-GFP. The cultures were harvested 48 h later, and the cell extracts were prepared according to Liu et al. (27). After 1 h of incubation with anti-GFP (25 g/ml) antibodies (Clontech), aliquots (200 l) from this mixture were incubated with 30 l of protein A-agarose (Invitrogen) at 4°C overnight. After washing five times with immunoprecipitation buffer (28), the bound proteins were released by boiling in 20 l of 2ϫ SDS loading buffer for 3 min (26). The released proteins were examined by Western blotting with sodium channel pan-specific antibodies (Upstate), and the signal was detected using the ECL chemiluminescent system (Amersham Pharmacia Biotech).
Expression of Recombinant Na ϩ Channels-HEK 293 cells were grown under culture conditions and co-transfected with 1 g of wild type human Na v 1.5 or its D1790G mutant cDNAs, together with 0.9 g of FHF1B-GFP and pEGFP-N1 cDNAs, using a calcium-phosphate protocol. Transfected cells were identified by green fluorescence from recombinant FHF1B-GFP and EGFP proteins.
Electrophysiology-Na v 1.5 sodium currents were measured by whole cell patch-clamp protocols (29) with Axopatch 200B amplifiers (Axon Instruments) using the following solutions: internal, 140 mmol/liter CsF, 1 mmol/liter EGTA, 10 mmol/liter NaCl, and 10 mmol/liter HEPES, pH 7.3, adjusted to 310 mOsmol/liter with glucose; external, 140 mmol/liter NaCl, 3 mmol/liter KCl, 1 mmol/liter MgCl 2 , 1 mmol/ liter CaCl 2 , and 20 mmol/liter HEPES, pH 7.3, adjusted to 320 mOsmol/ liter with glucose. The pipette potential was zeroed before seal formation; the voltages were not corrected for liquid junction potential. Capacity transients were cancelled, and series resistance was compensated (90%) as necessary. Leakage current was digitally subtracted on-line using hyperpolarizing control pulses, applied before the test pulse, of one-sixth test pulse amplitude (ϪP/6 procedure). Whole cell currents were elicited from a holding potential of Ϫ130 mV and filtered at 5 kHz and acquired at 50 kHz using Clampex 8.1 software (Axon Instruments). For current density measurements, the membrane currents were normalized to membrane capacitance, which was calculated as the integral of the transient current in response to a 5-ms hyperpolarizing pulse from Ϫ120 mV (holding potential) to Ϫ130 mV. All of the experiments were performed at room temperature (21-25°C).
Experimental Protocols and Data Analysis-Steady-state activation curves were constructed with the membrane potential held at Ϫ130 mV and a series of 100 ms test pulses that ranged from Ϫ100 to ϩ60 mV in increments of 5 mV. The peak value of I Na (I Na,peak ) at each membrane potential (V m ) was plotted. The relationship of peak I Na versus V m was fitted with the following equation where V rev is the reversal potential of I Na , and G Na is the voltage-dependent sodium current conductance. G Na was fitted using the following Boltzmann distribution equation where G Na,max is the maximum conductance, V1 ⁄2 is the membrane potential at half-maximal conductance, and k is the slope factor. The voltage-dependent steady-state inactivation relationship was investigated with a standard two-pulse protocol. For maximum recovery from slow inactivation of Na ϩ channels, a test potential to Ϫ20 mV was preceded by a 500-ms preconditioning pulse ranging from Ϫ130 to Ϫ10 mV in wild type and Ϫ150 to Ϫ10 mV in mutants. The normalized curves (I Na /I Na,max ) were fitted using the following Boltzmann distribution equation where I Na,max is the peak Na ϩ current at Ϫ20 mV measured from the most negative preconditioning pulse potential (Ϫ130 mV for wild type or Ϫ150 mV for mutant), V m is the preconditioning pulse potential, V1 ⁄2 is the membrane potential of half-maximum I Na , and k is the slope factor. The data are represented as the means Ϯ S.E.; a value of p Ͻ 0.05 was considered statistically significant.

FHF1B
Associates with Cardiac Channel Na v 1.5 and Sensory Neuron-specific Channel Na v 1.9 in Yeast Two-hybrid Assays-We reported previously that FHF1B binds directly to the C terminus of sensory neuron-specific, voltage-gated TTX-R sodium channel Na v 1.9 (23). The C-terminal polypeptides of sodium channels vary in length from 177 amino acids for Na v 1.9 to 251 amino acids for Na v 1.3a (brain type IIIA); alignment of these polypeptides (not shown) shows that the similarity to the C-terminal polypeptide of Na v 1.9 ranges from 50% for Na v 1.1a to 56.7% for Na v 1.5. Despite the 50% similarity of the C termini of the sodium channels, FHF1B did not bind to the C termini of Na v 1.7 and Na v 1.8, which are expressed, as is Na v 1.9, in adult rat dorsal root ganglia neurons (23). FHF1B is expressed in tissues where other sodium channels are expressed; therefore, we tested whether the C-terminal polypeptides of other sodium channels bind FHF1B in yeast two-hybrid assays. As shown in Fig. 1, FHF1B binds to the C-terminal polypeptide of Na v 1.5 and Na v 1.9 but not to other sodium channels in this assay.
FHF1B Associates with Na v 1.5 Both in Vitro and in Vivo-To test whether FHF1B binds to the C-terminal polypeptide of Na v 1.5 in vitro, we used the pull-down assay. We expressed the FHF1B-GFP fusion protein in HEK293 cells and the C-terminal polypeptide of Na v 1.5 as a GST fusion protein (GST-Na v 1.5C) in bacteria. Affinity-purified GST and GST-Na v 1.5C immobilized on glutathione-Sepharose beads were incubated with lysates of HEK293 cells expressing either GFP or FHF1B-GFP. Purified GST or GST-Na v 1.5C did not pull down GFP protein in this assay ( Fig. 2A, lanes 1-3). Purified GST did not pull down FHF1B-GFP fusion protein ( Fig. 2A, lane 5), whereas GST-Na v 1.5C efficiently pulled down the FHF1B-GFP fusion protein (Fig. 2A, lane 6).
To examine whether FHF1B also binds to full-length Na v 1.5 in vivo, we performed a co-immunoprecipitation assay (Fig.  2B). In this assay, HEK293 cells were transfected with fulllength human Na v 1.5 together with GFP (lanes 1 and 3) or Na v 1.5 together with full-length FHF1B-GFP (lane 2 and 4). Comparable levels of Na v 1.5 were detected in total lysates from both conditions using a pan-specific sodium channel antibody (lanes 1 and 2). Lysates from sister cultures were incubated with anti-GFP antibodies (lane 3 and 4), and the immunoprecipitated complexes were counter-tested with the sodium channel pan-specific antibodies. A single immunoreactive band of Na v 1.5 expected size is detected in the sample brought down by anti-GFP antibodies from the cell lysates expressing full-length Na v 1.5 together with FHF1B-GFP (lane 4) but not from the lysates expressing full-length Na v 1.5 with GFP (lane 3). Thus, FHF1B specifically binds to full-length Na v 1.5 in vivo.
The N-terminal Segment (a.a. 1-41) of FHF1B Is Required for Association with the C-terminal Polypeptide of Na v 1.5-The segment of FHF1B that is critical for the binding of the C terminus of Na v 1.5 was mapped by a pull-down assay using a series of N-terminal deletions of FHF1B. FHF1B derivatives were linked to GFP (Fig. 3A), and the constructs were used to transfect HEK 293 cells. Pull-down assays were performed using immobilized GST (Fig. 3B, lanes 2 , 5, 8, 11, 14, and 17) and GST-Na v 1.5C (Fig. 3B, lanes 3, 6, 9 , 12, 15, and 18), and cell extracts were prepared from transfected HEK293 cells. The cell lysates show the presence of the respective GFP-FHF1B derivative (Fig. 3B, lanes 1, 4, 7, 10, and 13) or GFP alone (Fig.  3B, lane 16). Immunoblotting using anti-GFP antibodies showed that deletion of the N-terminal tetrapeptide of FHF1B significantly reduced association with GST-Na v 1.5C compared with the full-length factor (compare lanes 3 and 6). Deletion of N-terminal 41 amino acid residues totally disrupted the inter- FIG. 1. FHF1B associates with Na v 1.5 and Na v 1.9. The ␤-galactosidase assay was used to test the interaction between FHF1B and the C termini of the voltage-gated sodium channels (rNa v 1.1, rNa v 1.2, rNa v 1.3, rNa v 1.4, rNa v 1.5, mNa v 1.6, hNa v 1.7, rNa v 1.8, and rNa v 1.9). The known interaction between c-Jun and c-Fos serves as a positive control, whereas the lack of interaction of retinoblastoma (Rb) and laminin serves as a negative control. Three independent yeast transformants for each pair of plasmids were transferred onto the nitrocellulose membrane, and the ␤-galactosidase activity was determined.

FIG. 2. FHF1B associates with Na v 1.5 both in vitro and in vivo.
A, binding of FHF1B to Na v 1.5C in vitro (GST pull-down assay). Purified GST (lanes 2 and 5) or GST-Na v 1.5C fusion protein (lanes 3 and 6) immobilized on glutathione-Sepharose beads were incubated with extracts prepared from HEK293 cells transfected with a plasmid encoding either GFP alone (lanes 1-3) or FHF1B-GFP fusion protein (lanes 4 -6). HEK293 proteins that were trapped by the interaction with GST or GST-Na v 1.5C (lanes 1 and 4, respectively) were examined by immunoblotting with anti-GFP antibodies. The positions of GFP and FHF1B-GFP are indicated. B, FHF1B binds to Na v 1.5 in vivo (co-immunoprecipitation assay). Equal amounts of protein extracts prepared from HEK293 cells transfected with Na v 1.5 expression plasmid with either pEGFP-N1 (lanes 1 and 3) or pFHF1B-GFP (lanes 2 and 4) were incubated with anti-GFP polyclonal antibodies (lanes 3 and 4). The immunoprecipitated protein complex and cell extracts (lanes 1 and 2), which provides a positive control) were examined by immunoblotting with a sodium channel pan-specific antibody. The position of Na v 1.5 is indicated.
action of FHF1B and Na v 1.5C (lane 9). FHF1B lacking the N-terminal 77 or 143 residues (lane 11 and 15), similar to GFP alone (lane 18), did not show any interaction with GST-Nav1.5C. As a negative control, GST did not pull down either of the FHF1B derivative polypeptides (see Fig. 6B, lanes 2, 5, 8,  11, 14, and 17). These data show that the N-terminal segment (a.a. 1-41) contains molecular determinants for the interaction of FHF1B with the C-terminal polypeptide of Na v 1.5.
Conserved Segments of the C-terminal Polypeptide of Na v 1.5 Are Important for FHF1B Binding-Alignment of the C-terminal polypeptides of the three TTX-R Na ϩ channels (Fig. 4) and all mammalian sodium channels (not shown) identified two conserved regions (CR1 and CR2) of 60 and 54 amino acids each and a terminal unique region of 132 amino acids for Na v 1.5. We subcloned fragments encoding the individual conserved and unique regions (Fig. 5A) to identify the Na v 1.5 segment, which is necessary for FHF1B binding. We carried out yeast twohybrid assays using CR1, CR2, CR1ϩCR2, and the unique region linked to GAL4DBD and full-length FHF1B fused toVP16AD. As shown in Fig. 5, removal of the unique region (a. a. 1885-2016) and CR2 (a.a. 1832-1885) from the C terminus did not disrupt the interaction with FHF1B. CR1 (a.a. 1773-1832) alone is necessary and sufficient to bind FHF1B.
LQT-3 Point Mutation in Na v 1.5 Abolishes Interaction with FHF1B-CR1 of Na v 1.5 C terminus (also referred to as the acidic rich domain) includes a number of negatively charged residues (Fig. 4), which is important in the control of the channel inactivation (30,31). Several inherited mutations that are associated with the cardiac arrhythmias Long QT syndrome and Brugada disease have been reported in this domain. The C-terminal polypeptide amino acid sequences of the channels are aligned using the Clustal W program. CR1 and CR2 are underlined. Residues in CR1 and CR2 that are conserved in Na v 1.5 and Na v 1.9 but that are different in Na v 1.8 are shown in bold type. The shaded sequences represent the predicted six helical segments in the C terminus of Na v 1.5 (31). The position of the cardiac arrythmia mutants E1784K (32), D1790G (33-35), and Y1795H and Y1795C (36) are shown. An asterisk represents identical amino acid residues in the three channels; a colon represents conservative substitutions; a dot represents a semi-conservative change; and a space represents a missing residue.  5. CR1 (a.a. 1773-1832) of the C-terminal polypeptide of Na v 1.5 is sufficient for FHF1B binding. A, schematic diagrams of DBD-Na v 1.5C fusion proteins used to map the sites of FHF1B binding on the C-terminal polypeptide of Na v 1.5. The numbers refer to amino acid residues in the C terminus of Na v 1.5. CR1 (green), CR2 (brown), and UR (magenta) stand for conserved region 1, conserved region 2, and unique region, respectively. The binding or lack of binding of FHF1B to the various DBD-Na v 1.5C segments, as revealed in B, are indicated by plus and minus signs, respectively. B, binding of FHF1B by the C terminus of Na v 1.5 and its derivatives. The ␤-galactosidase assay was used to test the interaction between FHF1B and derivatives of the C terminus of Na v 1.5. Three independent yeast transformants for each pair of plasmids were transferred onto the nitrocellulose membrane, and the ␤-galactosidase activity was determined as previously described.
The effect of the D1790G mutation on the interaction of FHF1B and full-length Na v 1.5 in vivo was tested by an immunoprecipitation assay (Fig. 6C). In this assay, extracts of cells expressing the mutant D1790G (lanes 1 and 2) or wild type human Na v 1.5 (lanes 3 and 4) together with full-length FHF1B-GFP were first incubated with anti-GFP antibodies (lane 2 and 4), and the immunoprecipitated complexes were counter-tested with sodium channel pan-specific antibodies. Both wild type Na v 1.5 and the D1790G mutant channel were expressed efficiently as comparable levels of the channels were detected in the cell lysates (lanes 1 and 3). Full-length wild type Na v 1.5 (Fig. 6C, lane 4) but not the D1790G mutant (lane 2) co-immunoprecipitates with FHF1B-GFP. Thus, FHF1B specifically binds to full-length Na v 1.5 in vivo, and the LQT-3 mutation D1790G abolishes this interaction.
Voltage-dependent Sodium Currents in HEK293 Cells Transfected with Na v 1.5 or Na v 1.5ϩFHF1B-The observations that FHF1B binds to the C-terminal polypeptide of Na v 1.5 and that anti-GFP antibody precipitates the full-length channel together with FHF1B-GFP, prompted us to investigate whether this binding modulates Na v 1.5 currents. For this purpose, Na v 1.5 channels were co-expressed together with FHF1B-GFP in HEK 293 cells and were compared with Na v 1.5 channels that were co-expressed with GFP alone. Voltage-gated Na ϩ currents with fast activation and fast inactivation kinetics were evoked by depolarizing pulses in cells transiently transfected with Na v 1.5 plus GFP or co-transfected with Na v 1.5 plus FHF1B-GFP (Fig. 7, A and B). Functional association of the FHF1B protein with Na v 1.5 channel did not significantly change the peak current densities. The peak Na ϩ current density (elicited by voltage pulses from Ϫ130 to Ϫ30 mV) was 0.47 Ϯ 0.11 nA/pF (n ϭ 24) for Na v 1.5 plus GFP and 0.28 Ϯ 0.03 nA/pF for Na v 1.5ϩFHF1B-GFP (n ϭ 22, p ϭ 0.18), respectively.
Na v 1.5 Channel Activation Is Not Modulated by FHF1B- Fig. 7C shows the current-voltage (I-V) relationship of the voltage-dependent activation of Na v 1.5 with or without FHF1B. Sodium currents were activated at around Ϫ60 mV and reached maximal amplitude at Ϫ30 mV for both Na v 1.5 (n ϭ 11) and Na v 1. 5ϩFHF1B (n ϭ 19). The general shape of the I-V relationship did not change. The midpoint of the normalized I-V curve (from Ϫ60 to Ϫ30 mV) for Na v 1.5ϩFHF1B showed a slight shift in the positive direction (Fig. 7D, open circles) but is not statistically significant (p ϭ 0.21). The average V1 ⁄2 and k (slope) values for the fitted functions were Ϫ50.41 Ϯ 1.32 mV and Ϫ5.6 Ϯ 0.47 mV/e-fold potential change, respectively, for Na v 1.5 and Ϫ46.24 Ϯ 1.13 mV and Ϫ5.8 Ϯ 0.44 mV/e-fold potential change, respectively, for Na v 1.5ϩFHF1B.
FHF1B Modulates Na v 1.5 Channel Inactivation-The effects of FHF1B on voltage-dependent inactivation were examined by FIG. 6. The LQT-3 point mutation D1790G in the C terminus of Na v 1.5 abolishes the interaction between FHF1B and Na v 1.5. A, schematic diagrams of wild type (Na v 1.5C), naturally occurring point mutants (E1784K, D1790G, Y1795H, and Y1795C) and combination mutant (KGH). The numbers refer to amino acid residues in the C terminus of Na v 1.5. The binding or lack of binding of FHF1B to the wild type and various point mutants, as revealed in B, are indicated by plus and minus signs, respectively. B, binding of FHF1B to the wild type and various mutants of the C terminus of Na v 1.5. The ␤-galactosidase assay was used to test the interaction between FHF1B and mutants of the C terminus of Na v 1.5. Three independent yeast transformants for each pair of plasmids were transferred onto the nitrocellulose membrane, and the ␤-galactosidase activity was determined. C, D1790G mutant abolishes the binding of FHF1B to Na v 1.5 in vivo (co-immunoprecipitation assays). Extracts prepared from HEK293 cells transfected with pFHF1B-GFP with either wild type Na v 1.5 (lanes 3 and 4) or D1790G mutant (lanes 1 and 2) expression plasmids were incubated with anti-GFP polyclonal antibodies (lanes 2 and 4). The immunoprecipitated protein complex and respective cell extract (lanes 1 and 3, which provides a positive control) were examined by immunoblotting with a sodium channel pan-specific antibody. IP, immunoprecipitation; CE, cell extract.

FIG. 7. Modulation of Na v 1.5 channel inactivation by FHF1B.
A, current traces recorded from HEK 293 cells co-transfected with Na v 1.5 and GFP. B, current traces recorded from HEK 293 cells cotransfected with FHF1B-GFP. C, normalized current-voltage relationships for Na v 1.5 (n ϭ 11) and Na v 1.5ϩFHF1B-GFP (n ϭ 19) are shown. D, voltage-dependent activation (normalized conductance, circles) and inactivation (fraction available, squares) were determined as described under "Experimental Procedures." The lines connecting the symbols are Boltzmann fits to the data. measuring the amplitude of peak currents evoked by a twopulse protocol as detailed under "Experimental Procedures." The average V1 ⁄2 of the voltage-dependent inactivation of Na v 1.5 was Ϫ76.40 Ϯ 1.2 mV with a k value of 8.4 Ϯ 0.5 mV/e-fold potential (n ϭ 16; Fig. 7D, open squares). In contrast, coexpression of FHF1B with Na v 1.5 caused a 9 mV hyperpolarizing shift of the V1 ⁄2 (Ϫ85.66 Ϯ 1.4 mV; p Ͻ 0.001; Fig. 7D, closed squares) but no change in the slope (k value of 9.1 Ϯ 0.2 mV/e-fold potential change (n ϭ 21)). Thus, the functional association of FHF1B with Na v 1.5 causes a significant hyperpolarizing shift in the voltage-dependent inactivation of the channel.
FHF1B Does Not Modulate D1790G Mutant Channel-Because LQT-3 point mutation D1790G abolishes the interaction between Na v 1.5 channel and FHF1B in yeast two-hybrid and biochemical assays (Fig. 6), we confirmed a lack of interaction with the mutant channel by electrophysiology. Voltage-gated Na ϩ currents were evoked by depolarizing pulses in HEK 293 cells transiently co-transfected with D7190G plus GFP or cotransfected with D1790G and FHF1B-GFP (Fig. 8, A and B). As seen in wild type Na v 1.5 currents, co-expression of FHF1B protein with the D1790G mutant channel did not significantly enhance the peak current density. The peak Na ϩ current densities (elicited by voltage pulses from Ϫ130 to Ϫ30 mV) were 0.17 Ϯ 0.06 nA/pF (n ϭ 17) for D1790G and 0.17 Ϯ 0.09 nA/pF for D1790GϩFHF1B (n ϭ 18, p Ͼ 0.05), respectively.
The Na ϩ currents produced by the D1790G mutant channel showed faster inactivation kinetics (Fig. 8A) compared with wild type Na v 1.5 channels, as reported earlier (35). The inactivation kinetics are 3-fold faster than the wild type Na v 1.5 currents at Ϫ30 mV, where maximal current amplitude is measured (data not shown). Co-expression of FHF1B did not alter the kinetics of inactivation of D1790G currents (Fig. 8B). The I-V relationships of D1790G and D1790GϩFHF1B were not significantly different (Fig. 8C). Voltage-dependent activation of D1790G and D1790GϩFHF1B currents are shown in Fig. 8D (circles). The average V1 ⁄2 and k values for the fitted functions were Ϫ40.27 Ϯ 3.15 mV and Ϫ8.68 Ϯ 0.59 mV/e-fold potential change, respectively, for D1790G (n ϭ 8) and Ϫ41.92 Ϯ 3.46 mV and Ϫ10.56 Ϯ 1.27 mV/e-fold potential change, respectively, for D1790GϩFHF1B (n ϭ 11). Compared with the wild type Na v 1.5 channels, the D1790G mutant channels show a depolarized shift of ϳ10 mV in voltage-dependent activation (p Ͻ 0.05) in agreement with published results (35). Co-expression of FHF1B and D1790G did not modulate the voltage-dependent activation of the mutant channels.
The D1790G mutation causes a marked negative shift, Ϫ22 mV, in the voltage dependence of inactivation (Fig. 8D, open squares; p Ͻ 0.001), in agreement with published results (35,37). The average V1 ⁄2 and k values for the fitted functions were Ϫ98.74 Ϯ 2.82 mV and 10.82 Ϯ 1.06 mV/e-fold potential change, respectively, for D1790G mutant (n ϭ 16). Co-expression of FHF1B with D1790G channels failed to affect voltagedependent inactivation of the channel. The average V1 ⁄2 and k values for the fitted functions were Ϫ98.43 Ϯ 2.91 mV and 10.33 Ϯ 1.7 mV/e-fold potential change, respectively, for D1790GϩFHF1B (n ϭ 16). The lack of modulation of D1790G channels by co-expression with FHF1B is consistent with the biochemical data showing that this mutation abolishes the binding of the proteins.

DISCUSSION
The results of this study show, for the first time, that interaction with a growth factor can modulate the properties of a voltage-gated sodium channel. Our findings show that a member of the FGF family (FGF12; also termed FHF1B) binds to two members of the voltage-gated sodium channel family (Na v 1.5 and Na v 1.9). The association of FHF1B with Na v 1.5 (this study) and with Na v 1.9 (23) were verified by both in vitro and in vivo assays. We show that the N-terminal 41 amino acids of FHF1B and the 60-amino acid CR1 (acidic rich domain, a.a. 1773-1832) in the proximal region of the C terminus of Na v 1.5 are important for the interaction of these two proteins.
Our results further demonstrate that binding of FHF1B to Na v 1.5 results in a significant alteration in the voltage dependence of inactivation. The LQT-3 channel mutant D1790G abolishes the binding of Na v 1.5 to FHF1B. Our findings thus demonstrate the binding of FHF1B to the C terminus of Na v 1.5 and provide the first evidence for the modulation of a sodium channel by a growth factor. FHF1B binds to the proximal segment of the C terminus of Na v 1.5 and modulates the voltage dependence of inactivation of the channel with no effects on channel activation (this study). Na v 1.5 is primarily expressed in heart muscle (38,39) but is also expressed in embryonic dorsal root ganglia neurons (40) and in limbic regions of rat brain (41). FHF1B also binds to the analogous region of the sensory neuron-specific TTX-R sodium channel Na v 1.9 (23). The effect of FHF1B interaction with Na v 1.9 has yet to be determined.
Na v 1.8 is a sensory neuron-specific sodium channel that produces a TTX-R current with different kinetic properties compared with Na v 1.5 and Na v 1.9 (42)(43)(44). Although Na v 1.5 and Na v 1.8 share ϳ80% amino acid identity in the CR1ϩCR2 region, Na v 1.8 failed to interact with FHF1B ( Fig. 1 and Ref. 23). In contrast, Na v 1.5 and Na v 1.9 share ϳ70% amino acid identity in the CR1ϩCR2 segment, and both bind FHF1B. Alignment of the C-terminal polypeptide of the three TTX-R channels (Fig. 4) identifies multiple residues in the CR1ϩCR2 region that are identical in Na v 1.5 and Na v 1.9 but that are different from Na v 1.8. It is possible that one or more of these residues may block or destabilize the interaction of FHF1B with the C terminus of Na v 1.8.
FHF1B lacks a signal peptide, similar to FGF2, and is not secreted into the medium by transfected HEK 293 cells (18 - A, current traces recorded from HEK 293 cells co-transfected with D1790G and GFP. B, current traces recorded from HEK 293 cells co-transfected with FHF1B-GFP. C, normalized current-voltage relationships for D1790G (n ϭ 8) and D1790GϩFHF1B (n ϭ 11) are shown. D, voltage-dependent activation (normalized conductance, circles) and inactivation (fraction available, squares) were determined as described under "Experimental Procedures." The lines connecting the symbols represent Boltzmann fits to the data. 21). FGF2 is exported to the extracellular space via an endoplasmic reticulum/Golgi-independent mechanism that may involve the ␣-subunit of Na ϩ /K ϩ -ATPase (45,46). We previously posited that Na v 1.9, in an analogous fashion to the ␣-subunit of Na ϩ /K ϩ -ATPase, might participate in the export of FHF1B to the cell surface in native neurons (23). Na v 1.9 produces only a very small current when expressed in HEK 293 cells (47) perhaps due, among other possibilities, to a low channel density at the cell surface. Thus, the lack of FHF1B in the medium of HEK 293 cells co-transfected with Na v 1.9 and FHF1B was not surprising. Therefore, because recombinant Na v 1.5 produces a robust current in HEK 293 cells and is modulated by FHF1B, indicating a functional interaction in the cell, we tested the FHF1B export hypothesis in the present study. Using anti-GFP antibodies, FHF1B-GFP was precipitated from HEK 293 cells co-transfected with Na v 1.5 and FHF1B-GFP but was undetectable in the medium (data not shown), suggesting that the Na v 1.5 channel does not enhance the export of FHF1B to the medium.
An early study speculated that the interaction between FHF1B and Na v 1.9 might modulate the channel properties either directly or indirectly by recruiting other proteins to the channel complex (23). FHF1 has been shown to interact with the scaffold protein IB2 and recruit mitogen-activated protein kinase p38␦ to the complex (22). The present study clearly shows that co-expression of FHF1B produces a significant hyperpolarizing shift of the voltage dependence of inactivation of Na v 1.5. This shift is similar to that attributed to the phosphorylation of Ser 1505 in the cytoplasmic linker joining domains 3 and 4 by protein kinase C (48,49). Other residues, for example Ser 526 and Ser 529 in the cytoplasmic loop linking domains 1 and 2 (L1) of Na v 1.5, have been reported to be phosphorylated by cAMP-dependent protein kinase (50). Na v 1.2 sodium channels are phosphorylated at serine residues in L1 by cAMP-dependent protein kinase anchored to the channel via the protein AKAP15 (13). Thus, it is possible that the binding of FHF1B to CR1 of the Na v 1.5 C terminus might recruit a kinase to the channel complex to phosphorylate either Ser 1505 or Ser 526 and Ser 529 or other residues and influence the inactivation of the channel. The identity of the kinase that might be recruited to Na v 1.5 and Na v 1.9 remains to be determined.
Sodium channels Na v 1.1 and Na v 1.5 and the auxiliary subunits ␤1, ␤2, and ␤3 are present in cardiac myocytes (39,51,52). The association of Na v 1.5 with ␤1 or ␤3 may produce a depolarizing shift in voltage dependence of inactivation (33,51,52) and in recovery from inactivation (51). Several inherited mutations linked to Long QT syndrome and Brugada syndrome have been reported in the acidic-rich domain in the C terminus of the cardiac Na v 1.5 channel, which overlaps with CR1, i.e. E1784K (32), D1790G (33)(34)(35), and Y1795H and Y1795C (36). An et al. (33) postulated that the D1790G mutation prevents the binding of ␤1 to Na v 1.5 and causes a significant shift in the opposite direction. Similarly, FHF1B may interfere with the binding of ␤ subunits to Na v 1.5. Co-expression of FHF1B and ␤1 subunit, however, did not affect the modulation of Na v 1.5 by FHF1B (data not shown), suggesting that the two proteins do not compete for the same binding site or that FHf1B binds to Na v 1.5 with higher affinity than the ␤-subunits.
The conserved acidic-rich domain in the C terminus of Na v 1.5 is important in the control of inactivation (30,31). Our findings that FHF1B binds to this domain and that it causes a significant shift in the voltage dependence of channel inactivation to more negative potentials support this concept. Our results also show that the D1790G mutation abolishes the interaction with FHF1B. If the FHF1B-mediated hyperpolarizing shift of the channel inactivation is caused by the recruit-ment of a kinase to the channel complex, the D1790G hyperpolarizing shift of channel inactivation might be produced by an alternate mechanism. Irrespective of this, the present results provide, for the first time, evidence that FHF1B, a member of the fibroblast growth factor family, can bind to and functionally modulate the cardiac sodium channel.
Because FHF1B binding to the Na v 1.5 mimics the effects of D1790G mutation, it is possible that the level of expression of FHF1B can affect heart function. Accordingly, the low level of FHF1B expression in adult heart (23) may be functionally important. Whether overexpression of FHF1B in the heart under pathological conditions or by transgenic techniques produces physiological changes similar to those of the Long QT syndrome needs to be further explored.