Cloning, Localization, and Functional Expression of Sodium Channel β1A Subunits*

Auxiliary β1 subunits of voltage-gated sodium channels have been shown to be cell adhesion molecules of the Ig superfamily. Co-expression of α and β1 subunits modulates channel gating as well as plasma membrane expression levels. We have cloned, sequenced, and expressed a splice variant of β1, termed β1A, that results from an apparent intron retention event. β1 and β1A are structurally homologous proteins with type I membrane topology; however, they contain little to no amino acid homology beyond the shared Ig loop region. β1A mRNA expression is developmentally regulated in rat brain such that it is complementary to β1. β1A mRNA is expressed during embryonic development, and then its expression becomes undetectable after birth, concomitant with the onset of β1 expression. In contrast, β1A mRNA is expressed in adult adrenal gland and heart. Western blot analysis revealed β1A protein expression in heart, skeletal muscle, and adrenal gland but not in adult brain or spinal cord. Immunocytochemical analysis of β1A expression revealed selective expression in brain and spinal cord neurons, with high expression in heart and all dorsal root ganglia neurons. Co-expression of αIIA and β1A subunits in Chinese hamster lung 1610 cells results in a 2.5-fold increase in sodium current density compared with cells expressing αIIA alone. This increase in current density reflected two effects of β1A: 1) an increase in the proportion of cells expressing detectable sodium currents and 2) an increase in the level of functional sodium channels in expressing cells. [3H]Saxitoxin binding analysis revealed a 4-fold increase in B max with no change inK D in cells coexpressing αIIA and β1A compared with cells expressing αIIA alone. β1A-expressing cell lines also revealed subtle differences in sodium channel activation and inactivation. These effects of β1A subunits on sodium channel function may be physiologically important events in the development of excitable cells.

Sodium channels isolated from brain are composed of a central pore-forming ␣ subunit and two auxiliary subunits, ␤1 and ␤2, which do not form the pore yet play critical roles in channel modulation and expression. A mutation in the ␤1 gene (SCN1B) has been implicated to play a role in febrile seizures and generalized epilepsy, GEFSϩ (1). The primary structure of the ␤1 subunit predicts an integral membrane glycoprotein with type I transmembrane topology as well as an extracellular Ig fold (2,3). ␤ 1 subunits can be classified as members of the V-set of the Ig superfamily, which includes many cell adhesion molecules. ␤1 and ␣ subunit co-expression has been well characterized in Xenopus oocytes and in mammalian cells. In oocytes, co-expression of type IIA (SCN2A) or I (SCN4A) ␣ subunits with ␤1 increases the proportion of sodium channels that function in a fast gating mode, accelerates the macroscopic rates of activation and inactivation, shifts the voltage dependence of inactivation in the hyperpolarizing direction, and increases the peak current amplitude consistent with increases in channel expression (4 -9). In Chinese hamster lung (CHL) 1 cells, stable coexpression of ␤1 with ␣IIA results in increased channel expression levels at the plasma membrane as well as moderate hyperpolarizing shifts in the voltage dependence of channel activation and inactivation (10).
␤1 mRNA is expressed only after birth in the developing brain (5,11). However, previous studies showing the developmental time course of ␤1 protein expression in rat forebrain suggested that multiple ␤1 subunit isoforms may be present (12). A 26-kDa ␤1-immunoreactive protein was observed at embryonic day 18. This protein was also expressed in adult adrenal gland, heart, skeletal muscle, and PC12 cells. After birth, there was a dramatic decrease in the level of this protein in brain, and little if any remained by postnatal day 14. The expression time course of this immunoreactive protein was complementary to that of ␤1 mRNA. Day 18 embryonic brain membranes also exhibited a low level of an immunoreactive peptide that migrated with an apparent molecular mass greater than 42 kDa. This protein was not detected in rat brain after birth. Other excitable tissues expressed multiple size forms of immunoreactive ␤1-like subunits as well. Adult rat heart and skeletal muscle membrane preparations exhibited 38-and 41-kDa bands on Western blots in addition to the 26-kDa band. The 41-kDa immunoreactive band observed in these studies was shown to be the adult rat brain isoform and was later identified as C1Aa.␤1 (4). The immunoreactive peptides identified in the previous study were detected with a polyclonal antibody raised against purified ␤1 subunits; thus, they could represent ␤1 subunit isoforms that contained signif-* This work was supported by a Johnson and Johnson Focused Giving Award and American Heart Association/Michigan Affiliate awards 35GB967 and 14GS978 (to L. L. I.), and Medical Research Council of Canada Grant MT-13485 (to D. S. R.). Use of the Genetics Computer Group Software Package was supported by the General Clinical Research Center Grant M01 RR00042 at the University of Michigan. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
The nucleotide sequence(s) reported in this paper has been submitted to the GenBank TM /EBI Data Bank with accession number(s) AF182949.
To test this hypothesis, we screened a rat adrenal cDNA library with a probe encoding only the coding region of ␤1. The present study reports the molecular cloning and functional expression of ␤1A, a splice variant of the ␤1 gene that contains identical amino-terminal and extracellular Ig fold regions as ␤1 followed by a significantly different extracellular juxtamembrane domain, predicted transmembrane region, and predicted intracellular COOH-terminal domain. ␤1A mRNA is expressed early in embryonic brain development and then disappears after birth. Western blot analysis of membrane preparations using an antibody to a unique, extracellular region of ␤1A not found in ␤1 showed that ␤1A protein is expressed in adult rat heart, skeletal muscle, and adrenal gland but was not detected in adult rat brain or spinal cord. Immunocytochemical analysis of ␤1A expression in adult rat tissues revealed high expression in heart and dorsal root ganglion and selective expression in some areas of the brain and spinal cord. ␤1A functions to increase channel expression at the plasma membrane when coexpressed with ␣IIA subunits in CHL fibroblasts. Unlike ␤1, however, mean steady state inactivation curves for ␣␤1A-expressing cell lines were shifted to more positive potentials than the mean inactivation curves for cells expressing ␣ alone. Previous studies showed that coexpression of ␣ and ␤1 subunits in CHL cells shifted the voltage dependence of inactivation to more negative potentials compared with ␣ alone (10). Therefore, the novel, carboxyl-terminal domains of ␤1A may be important for electrophysiological function. It has been shown previously that the extracellular domain of ␤1 is essential for expression and function of the ␣␤1 complex in Xenopus oocytes (13,14). We propose that the extracellular Ig fold, common to ␤1 and ␤1A, is essential for the observed increases in channel expression levels. Thus, this report introduces a novel splice variant of ␤1, ␤1A, and adds to our understanding of ␤1 structure-function relationships in terms of channel expression levels and electrophysiology.

EXPERIMENTAL PROCEDURES
Library Screening-A cDNA probe encoding nucleotides 345-911 of p␤1.C1Aa (4) was labeled with digoxigenin following the manufacturer's instructions (Roche Molecular Biochemicals) and used to screen a Zap Express rat adrenal cDNA library prepared by Stratagene (La Jolla, CA). pBK plasmids containing cDNA inserts that hybridized strongly to the probe were rescued from the phage according to the manufacturer's instructions, confirmed by Southern blot analysis, and sequenced using ThermoSequenase (Amersham Pharmacia Biotech).
Reverse Transcriptase-PCR from Rat Adrenal RNA-To confirm independently that the ␤1A transcript identified by library screening was expressed by rat adrenal gland, we amplified a region of ␤1A from the amino terminus past the region in which the amino acid sequence changed from identity to nonidentity to ␤1, or the putative splice site, by reverse transcriptase-PCR using adult rat adrenal gland total RNA as template and ␤1A-3 (5Ј-GAAGATGAGCGCTTTGAGG-3Ј, primer sequence common to ␤1 and ␤1A) and ␤1A-5 (5Ј-GAGAGACACAGCAA-GC, primer sequence unique to ␤1A) as oligonucleotide forward and reverse primers, respectively. Rat adrenal gland cDNA was synthesized from total RNA using Superscript II (Life Technologies, Inc.) according to the manufacturer's instructions in a total volume of 20 l. 2.3 g of total rat adrenal RNA (purified using Trizol reagent; Life Technologies) was used in the reaction. The PCR conditions were as follows: 1 l of cDNA, 0.5 M concentration of each primer, 200 M concentration of each dNTP (Roche Molecular Biochemicals), 5 l of Mg 2ϩ -free 10ϫ PCR buffer (Perkin-Elmer), and 1.5 mM MgCl 2 were mixed in a total volume of 50 l. Following a hot start at 94°C, 0.25 l of AmpliTaq DNA polymerase (Perkin Elmer) were added to the reaction tube, and the amplification cycle was started. The cycling parameters were as follows: 40 cycles of 45 s at 94°C, 20 s at 60°C, 90 s at 72°C. This was followed by 10 min at 72°C and then 4°C until the tubes were removed from the thermocycler (GeneAmp 2400, Perkin-Elmer). Analysis of the PCR products on a 1% agarose gel revealed a 750-base pair band (data not shown). The band was excised form the gel, subcloned into pCR2.1 (Invitrogen, Carlsbad, CA), and analyzed using ThermoSequenase (Amersham Pharmacia Biotech). The sequence obtained from this PCR clone was identical to that obtained from the original ␤1A clone plaquepurified from the adrenal cDNA library.
Rat ␤1 Gene-Intron 3 of the rat ␤1 gene (15) was amplified by PCR using rat genomic DNA as the template; oligonucleotides that encode ␤1 coding sequence flanking intron 3, VVDK (5Ј-AGATCCACCTGGAGG-TGGTGGACAAGG-3Ј) and ANRD (5Ј-ACACGATGGATGCCATATCT-CTGTTGG-3Ј), as forward and reverse primer, respectively; and the Expand Long Template PCR System (Roche Molecular Biochemicals). All oligonucleotide primers were synthesized by Life Technologies. The amplification conditions were as follows. 300 ng of rat genomic DNA (CLONTECH Laboratories, Inc., Palo Alto, CA), 250 ng of each primer, 1 mM concentration of each dNTP (Roche Molecular Biochemicals), and 5 l of Expand Buffer 3 were mixed in a total reaction volume of 25 l. Following a hot start at 95°C, 0.5 l of Expand DNA polymerase were added, and the amplification cycle was started. 40 cycles of the following regimen were performed: 94°C for 10 s, 58°C for 30 s, and 68°C for 4 min plus 20 s added to each successive cycle. The samples were then held at 4°C until removal from the thermocycler (GeneAmp 2400, Perkin-Elmer). The 5-kilobase pair PCR product (data not shown) was gel-purified and sequenced directly using oligonucleotide VVDK as the sequencing primer.
RNase Protection Analysis-A plasmid containing a RNase protection probe template (pRPA-1) was constructed corresponding to nucleotides 364 -533 in the ␤1A sequence. Briefly, a 169-nucleotide AluI/AccI fragment was excised from pBK.␤1A and ligated into the SmaI and AccI sites of pBluescript (Stratagene). The resulting plasmid was then sequenced using ThermoSequenase. To synthesize labeled cRNA, a 10-g aliquot of pRPA-1 was linearized with XhoI, ethanol-precipitated, resuspended in RNase-free water, and labeled with digoxigenin using the T3 MAXIscript kit (Ambion, Austin, TX) according to the manufacturer's instructions. Following a 2-h incubation at 37°C, the reaction was incubated at 95°C for 2 min, chilled on ice, and then treated with RNase-free DNase (2 units) for 15 min at 37°C. EDTA (final concentration 30 mM) was added to stop the reaction. Free nucleotides were removed by ethanol precipitation with 0.5 M ammonium acetate, and the final pellet was resuspended in 20 l of RNase-free water. The probe (RPA-1) was quantitated by comparison of serial dilutions of the labeled probe with serial dilutions of control digoxigenin-labeled RNA following the manufacturer's instructions. Typical labeled probed concentrations were 10 ng/l throughout our experiments.
RNase protection experiments were performed using the HybSpeed RPA kit (Ambion). Briefly, 20 g of rat embryonic day 18 brain RNA were mixed with 1 l of digoxigenin-labeled RPA-1 probe and 30 g of yeast tRNA in 0.5 M ammonium acetate plus 2.5 volumes of ethanol. The reaction tubes were left at Ϫ20°C for 15 min, and the RNA was precipitated by centrifugation in a microcentrifuge at top speed. The RNA was resuspended in 10 l of HybSpeed hybridization buffer that had been preheated to 95°C and vortexed vigorously, and the tubes were placed at 95°C for 3 min. The samples were then hybridized for 10 min at 68°C and digested with a mixture of RNase A and T1 (10 units/ml RNase A and 400 units/ml RNase T1) for 30 min at 37°C. 150 l of HybSpeed Inactivation/Precipitation mix were added to each reaction, and the RNA was precipitated and resuspended in 10 l of gel loading buffer 1. The reactions were separated on a 1.5-mm-thick 6% acrylamide TBE denaturing gel containing 7 M urea in the Mini-Protean gel format (Bio-Rad), transferred to nylon (Roche Molecular Biochemicals), and UV-cross-linked using a Stratalinker (Stratagene). Hybridization of the digoxigenin-labeled probe was detected with alkaline phosphatase-conjugated anti-digoxigenin Fab fragments (1:10,000 dilution) and CSPD chemiluminescent substrate solution (Roche Molecular Biochemicals) according to the manufacturer's instructions.
Preparation of RNA and Northern Blot Analysis-Time-mated pregnant female Harlan Sprague Dawley rats were anesthetized with 60 mg/kg Beuthanasia-D intraperitoneal (Schering-Plow Animal Health Corp., Kenilworth, NJ), and the fetuses were surgically removed. Embryonic day 9 rats were homogenized in their entirety in Trizol reagent (Life Technologies) to purify total RNA according to the manufacturer's instructions. Whole fetal brains were dissected at the remaining embryonic time points, and total RNA was purified using Trizol reagent. RNA was also subsequently prepared from the brains and adrenal glands of the adult female rats. Postnatal rats at the indicated ages were anesthetized with Beuthanasia-D, brains were dissected, and total RNA was purified with Trizol reagent. Northern blot analysis of 20 g of each sample was performed as described previously using a digoxigenin-labeled ␤1A antisense cRNA probe encoding nucleotides 428 -850 or a digoxigenin-labeled antisense cRNA probe specific to the 3Ј-un-translated region of ␤1 (nucleotides 1053-1508 of p␤1.C1Aa (4).
Construction of ␤1A Expression Vector-A plasmid containing ␤1 cDNA including an in-frame amino-terminal hemagglutinin (HA) epitope tag was obtained as a generous gift from the laboratory of R. A. Maue (Dartmouth University) (16). This construct has been shown to express functional ␤1 subunits in Xenopus oocytes. The HA-tagged ␤1 cDNA was recloned into the EcoRI and NotI sites of the mammalian expression vector pCIneo (Promega, Madison, WI) to create pCI.␤1-HA. pCI.␤1-HA was subsequently digested with AccI and NotI and agarose gel-purified to remove the 3Ј end of ␤1. The AccI restriction endonuclease site is common to ␤1 and ␤1A. pBK.␤1A cDNA was digested with AccI and NotI and gel-purified. The 3Ј end of ␤1A was then ligated into AccI/NotI-digested pCI.␤1-HA to create pCI.␤1A-HA. The junctions were then sequenced to confirm that the segments of ␤1 and ␤1A were successfully ligated in frame.
Transfection of SNaIIA Cells with HA-tagged ␤1A-SNaIIA cells were transfected with pCI.␤1A-HA using DOTAP as described previously (10). Because SNaIIA cells are resistant to G418 as a result of the original transfection of the ␣IIA subunit, pCI.␤1A-HA was cotransfected with pSV2*Hyg to confer resistance to the antibiotic hygromycin. Drug selection with hygromycin (400 g/ml) required approximately 1 week; clonal cell lines were selected, analyzed by Northern blot, and expanded for further analysis. Antibodies-A multiple antigenic peptide with amino acid sequence RWRDRWKEGDRLVSHRGQ, encoded by nucleotides 160 -177 of ␤1A, was synthesized by the Protein and Carbohydrate Structure Facility at the University of Michigan. Rabbit polyclonal antibodies were subsequently generated in two separate animals and tested by enzyme-linked immunosorbent assay against the ␤1A peptide to determine the antibody titer (Research Genetics, Inc., Huntsville, AL).
Western Blot Analysis of ␤1A Protein Expression-Adult female Harlan Sprague Dawley rats were sacrificed by decapitation. Brain, spinal cord, heart, skeletal muscle, and adrenal gland tissues were immediately removed, minced, and briefly stored on ice. SNa␤1A-16 cell line cells were washed with PBS and scraped into 50-ml conical tubes. Membranes were prepared as described previously (10), and the final pellets were resuspended in 50 mM Tris, pH 8, 10 mM EGTA containing Complete-Mini protease inhibitor tablets according to the manufacturer's instructions (Roche Molecular Biochemicals). The total protein in each membrane preparation was quantitated with the BCA Protein Assay Kit (Pierce) using bovine serum albumin as the standard. 250 g of each membrane preparation were separated by SDS-PAGE as described previously (10), transferred to nitrocellulose (HyBond ECL; Amersham Pharmacia Biotech), and stained with Ponceau-S prior to immunodetection. Western blot analysis was performed as follows. The blot was washed for 10 min in TBS-T (10 mM Tris, pH 7.4, 150 mM NaCl, 0.1% Tween 20) at room temperature and then blocked for 1 h in 5% nonfat dry milk in TBS-T at room temperature. Primary anti-␤1A antibody (1:750 dilution) was applied in blocking solution for 30 min at room temperature. The blot was then washed five times for 15 min each in TBS-T. Secondary antibody (horseradish peroxidase-conjugated goat anti-rabbit IgG, ICN) diluted to 1:100,000 in blocking solution was applied for 30 min at room temperature. The blot was then washed five times for 15 min each in TBS-T. SuperSignal WestFemto chemiluminescent substrate solution (Pierce) was applied according to the manufacturer's instructions, and the blot was placed between plastic sheet protectors and exposed to Hyperfilm-ECL (Amersham Pharmacia Biotech) for the indicated times (typically 10 -30 s) at room temperature.
Immunohistochemical Analysis of ␤1A Expression-Sprague Dawley rats (4 -6 weeks of age; Harlan Industries, Indianapolis, IN) were perfused with 4.0% neutral buffered formalin for approximately 20 min, and tissues were removed for further processing. Tissues were postfixed overnight in 10% neutral buffered formalin, processed, embedded in paraffin blocks, and sectioned onto slides (5-m thickness). Tissues were processed for immunohistochemistry as described previously (17). Briefly, slides were blocked with normal goat serum, incubated with rabbit anti-rat ␤1A antibody at a dilution of 1:600 and then incubated with biotinylated goat anti-rabbit IgG (Vector Labs, Burlingham, CA). All incubation were performed for 30 min at room temperature. Detec-tion was accomplished using the ABC-horseradish peroxidase system (Vector Labs) followed by 3Ј-diaminobenzidine (Biomeda, Foster City, CA) as the chromogen, stained in Mayer's hematoxylin, and coverslipped with Permount (Fisher).
Electrophysiological Analysis-Electrophysiological recordings on SNaIIA and SNaIIA␤1A cells were performed by the patch clamp technique in the whole cell configuration (18), using an Axopatch 200B patch clamp amplifier and pCLAMP software (Axon Instruments, Foster City, CA). Data were filtered at 5 kHz and digitally sampled at 50 kHz. Series resistance was compensated 60 -80%. Capacitive transients, elicited by voltage steps, were partially canceled using the internal clamp circuitry. Additional subtraction of transients and leak currents was obtained using the P/4 procedure (19). For whole cell recordings, recording pipettes were filled with 105 mM CsF, 10 mM CsCl, 10 mM NaCl, 10 mM EGTA, 10 mM HEPES, pH 7.4, with CsOH. Pipette resistances were 1-3 megaohms. The bath solution consisted of 130 mM NaCl, 4 mM KCl, 1.5 mM CaCl 2 , 1 mM MgCl 2 , 5 mM glucose, 10 mM HEPES, pH 7.4, with NaOH. As described previously (10), the voltage dependence of sodium current activation and inactivation progressively shifted to more negative potentials over the first few minutes of experiments with fluoride-based intracellular solutions. Thus, all experiments were begun 10 min after break in, at which point the shifts in channel gating had stabilized.
For each cell, we examined the voltage dependence of current activation and steady state inactivation. Activation was assessed by applying test pulses to potentials from Ϫ50 to ϩ70 mV in 5-mV steps, following a 100-ms prepulse to Ϫ100 mV. Peak current amplitude (I peak ) was measured at each test potential and converted to conductance (g) according to the equation, g ϭ I peak /(V rev Ϫ V test ), in which V test is the test potential and V rev is the current reversal potential, determined by linear extrapolation of the straight line portion of the falling phase of the current-voltage relationship. The conductance values were normalized with respect to the maximal conductance, plotted as a function of V test , and fit with the Boltzman equation which V is the midpoint of the curve and k is a slope factor. Steady state inactivation was examined by applying 100-ms-long prepulses to potentials ranging from Ϫ100 to Ϫ10, in 5-mV steps, followed by a test pulse to 0 mV. The peak amplitude of currents evoked by the test pulses were normalized with respect to the largest currents, plotted as a function of prepulse potential, and fit with the Boltzman equation.

Molecular
Cloning and Analysis of ␤1A -A rat adrenal gland cDNA library prepared in the Zap Express vector was screened with a digoxigenin-labeled cDNA probe encoding nucleotides 345-911 of p␤1.C1Aa (4). A clone encoding a protein with a 5Ј region of identity to ␤1 and a novel 3Ј region was identified by DNA sequencing. The identity of this clone was then confirmed independently by reverse transcriptase-PCR from rat adrenal cDNA using the oligonucleotides ␤1A-3 and ␤1A-5 followed by DNA sequencing, as described under "Experimental Procedures." This clone, designated ␤1A, encoded a novel 253-amino acid protein of 29,055 daltons (predicted molecular mass of the mature protein with the signal sequence removed), which contains a predicted amino-terminal region of identity to ␤1, residues Met Ϫ1 through Lys 129 , followed by a novel carboxyl-terminal region (Fig. 1, A and B). Hydrophobicity analysis of the novel, carboxyl-terminal region revealed an apparent 66-amino acid extension of the extracellular region of ␤1 followed by a 19-amino acid transmembrane domain and short, 39-amino acid intracellular carboxyl terminus (Fig. 1, A  and C). The novel 3Ј region of ␤1A is structurally homologous to ␤1 in that it predicts a transmembrane domain and short intracellular region, yet it contains little to no homology at the amino acid level (Fig. 1B). Interestingly, the amino-terminal region common to ␤1 and ␤1A contains the extracellular immunoglobulin fold. ␤1A can thus be characterized structurally as a cell adhesion molecule (17).
Analysis of the novel 3Ј region of ␤1A by BLAST-P search of the Swiss-Prot data base revealed a 55-residue region of ␤1A with 32% identity to an extracellular low density lipoprotein receptor class A domain of human low density lipoprotein receptor-related protein 2 (LRP2), also called megalin or glyco-protein 330 (Fig. 1D) (20 -23). This region of homology in ␤1A is predicted to be located extracellularly, just proximal to the plasma membrane followed by the transmembrane region. LRP2 has been shown to be a cysteine-rich type I membrane protein that forms a multimeric complex with receptor-associated protein. LRP2 binds clusterin with high affinity and is localized to clathrin-coated pits, suggesting that it may be an endocytic receptor. Interestingly, LRP2 has been shown to interact with extracellular matrix components, similar to sodium channel ␤1 and ␤2 subunits (24). The BLAST-P search also revealed a 63-residue region of ␤1A with 26% identity to tensin, a protein that has been implicated as the anchor for actin filaments at focal adhesions and is thought to act as a link between the cytoskeleton and signal transduction proteins (25). The region of homology to ␤1A is located in the insertion domain of tensin. This domain has been shown to permit polymerization of actin filaments.
␤1A Is Encoded by a Retained Intron in the ␤1 Gene-The genomic organization of the human sodium channel ␤1 subunit gene has been reported previously (15). Using this information, we determined that the site of divergence between the ␤1 and ␤1A cDNAs was located precisely at the exon 3-intron 3 boundary of the ␤1 gene (Fig. 1E). Furthermore, a consensus sequence for exon-intron boundaries in genomic DNA was readily identified at this location. Amplification of intron 3 (approximately 5 kilobase pairs; data not shown) from rat genomic DNA by PCR followed by sequencing showed that the sequence of ␤1A beyond the amino acid sequence VVDK was indeed that of intron 3. We next performed a series of RNase protection experiments using a probe that was designed to span the exon 3-intron 3 boundary in the rat sequence. Fig. 2 shows that this 169-nucleotide probe was fully protected by rat embryonic day 18 brain mRNA. Thus, we propose that the novel extracellular, transmembrane, and carboxyl-terminal domains of ␤1A are encoded by alternative splicing of a retained intron within the ␤1 gene that includes an in-frame termination codon. These data are in agreement with the previously reported observation that ␤1 is represented only once in the rat and human genomes (26,27).
␤1A mRNA Is Expressed in Embryonic Brain and Adult Adrenal Gland-A comparison of the developmental time courses of ␤1A and ␤1 mRNA expression in developing rat ␤1A is predicted to encode a type I membrane protein with similar topology to ␤1. The disulfide-linked immunoglobulin fold is common to both ␤1 and ␤1A. Following the alternate splice site (arrow), ␤1A contains a novel 66-amino acid juxtamembrane region, 19-amino acid transmembrane segment, and 39-amino acid intracellular domain. C, sequence homology to known proteins. Results of BLAST-P search of the Swiss-Prot data base using the novel, carboxyl-terminal domain of ␤1A beginning at residue 130 as the query sequence. Accession numbers for LRP2 and tensin are indicated. D, schematic structure of the ␤1/␤1A gene. As reported previously, the ␤1 gene contains five introns, indicated as I1 through I5, and six exons. The ␤1A cDNA is the result of retention of I3, creating a novel 3Ј end. brain was determined using specific, noncross-hybridizing antisense cRNA probes for ␤1A and ␤1, respectively. Fig. 3 compares the developmental time course of expression of ␤1A (upper panel) versus ␤1 (lower panel) in total rat brain RNA from embryonic day 9 through postnatal day 21. The transcript size of ␤1A is approximately 4.4 kilobase pairs and reflects the retention of a portion of intron 3. Interestingly, the expression time course of ␤1A parallels that of the 26-kDa ␤1-immunoreactive band described previously (12) and complements the expression pattern of ␤1 (5) Thus, ␤1A is expressed early in development and disappears after birth. In contrast, ␤1 expression is not detectable during embryonic brain development and becomes detectable as ␤1A mRNA expression is decreased.
Analysis of ␤1A Protein Expression-To determine whether alternative splicing of the ␤1 gene resulted in expression of a novel protein, a polyclonal antibody was generated against a multiple antigenic peptide containing the amino acid sequence RWRDRWKEGDRLVSHRGQ encoded by the retained portion of intron 3 found in the ␤1A cDNA clone. Aliquots of membrane preparations from adult rat brain, heart, skeletal muscle, spinal cord, and adrenal gland were separated by SDS-polyacrylamide gel electrophoresis and blotted to nitrocellulose, and Western blot analysis was performed using the ␤1A antibody described above (1:750 dilution). As shown in Fig. 4, ␤1A immunoreactive bands migrating at approximately 45-50 kDa were observed in heart, skeletal muscle, and adrenal gland but were not detected in brain or spinal cord. An immunoreactive doublet was observed in adrenal gland. The absence of immunoreactive ␤1A protein bands in adult brain and spinal cord is consistent with the Northern blot results shown above.
Immunohistochemical Analysis of ␤1A Expression-Positive ␤1A protein immunoreactivity was detected as brown product (Figs. 5 and 6). Product was not detected in the negative controls, which included the replacement of the primary antibody with similar species isotype serum (data not shown). Fig. 5 represents various central and peripheral neuronal populations that contained ␤1A. Fig. 5, A and B, demonstrates that ␤1A was found in most but not all Purkinje cells in the cerebellum. In addition, some ␤1A-positive neurons in the dentate nucleus of the cerebellum were observed (Fig. 5C). ␤1A was absent from the granular layer (G) and the molecular layer (M) of the cerebellum. In Fig. 5D, a distinct population of pyramidal neurons of the cerebral cortex contained ␤1A, while glial cell populations remained negative. Spinal cord also contained several distinct populations of ␤1A-containing neurons. Fig. 5E shows ␤1A in a small population of motor neurons, while ␤1A rats were anesthetized with 60 mg/kg Beuthanasia-D intraperitoneally, and the fetuses were surgically removed. Embryonic day 9 fetuses were homogenized in their entirety in Trizol reagent to purify total RNA according to the manufacturer's instructions. Whole fetal brains were dissected at the remaining embryonic time points, and total RNA was purified using Trizol reagent. RNA was also prepared from the brains and adrenal glands of the adult female rats. Postnatal rats at the indicated ages were anesthetized with Beuthanasia-D, brains were dissected, and total RNA was purified with Trizol reagent. Northern blot analysis of 20 g of each sample was performed using a digoxigenin-labeled ␤1A antisense RNA probe encoding nucleotides 428 -850 or a digoxigenin-labeled antisense RNA probe specific to the 3Ј-untranslated region of ␤1 (nucleotides 1053-1508 of p␤1.C1Aa (4) A control reaction contained embryonic brain RNA and yeast tRNA with no added probe. The reaction tubes were left at Ϫ20°C for 15 min, and the RNA was precipitated by centrifugation in a microcentrifuge at top speed. The RNA was resuspended in 10 l of HybSpeed hybridization buffer that had been preheated to 95°C and vortexed vigorously, and the tubes were placed at 95°C for 3 min. The samples were then hybridized for 10 min at 68°C and digested with a mixture of RNase A and T1 (10 units/ml of RNase A and 400 units/ml RNase T1) for 30 min at 37°C. 150 l of HybSpeed inactivation/precipitation mix were added to each reaction, and the RNA was precipitated and resuspended in 10 l of gel loading buffer 1. The reactions were electrophoresed on a 1.5-mm-thick 6% acrylamide TBE denaturing gel containing 7 M urea in the Mini-Protean gel format, transferred to nylon, and UV-cross-linked. Hybridization of the digoxigenin-labeled probe was detected with alkaline phosphatase-conjugated anti-digoxigenin Fab fragments (1:10,000 dilution) and CSPD. The blot was then exposed to Hyperfilm-ECL at room temperature. Lane 1, embryonic brain RNA plus RPA-1 probe digested with RNase mixture; lane 2, embryonic brain RNA digested with RNase mixture; lane 3, undigested RPA-1 probe; lane 4, RNA ladder. The lower diagram indicates predicted results of undigested, fully protected, and partially digested probes.
neurons were also observed in laminae II-V of the dorsal horn. All neuronal cell types of the dorsal root ganglia contained ␤1A (Fig. 5F), while processes and glial cells were negative for the ␤1A product. Fig. 6, A-D, shows examples of nonneuronal sites of ␤1A localization. In Fig. 6A, the membranes of individual muscle fibers in rat atria were positive for ␤1A. Additionally, other areas of the rat heart such as ventricles contained ␤1A product (data not presented). ␤1A immunoreactivity was also observed in the alveoli (Fig. 6B) and in some bronchus columnar epithelial cells (Fig. 6C). Finally, labeling of endothelial cells in the pulmonary artery is shown in Fig. 6D.
Mammalian Cell Expression-To investigate the functional role of ␤1A, we constructed a HA epitope-tagged version of the ␤1A cDNA. We included the epitope tag for potential use in the event that our polyclonal antibody production was unsuccessful. We were successful in raising an anti-␤1A antibody and thus did not use the HA tag in these experiments. We created stably transfected cell lines expressing ␤1A in the previously characterized SNaIIA cell line. SNaIIA cells are a stable line expressing type IIA sodium channel ␣ subunits in CHL cells (10). Because SNaIIA cells are G418-resistant, we co-transfected pcDNA3-␤1A with pSV2*Hyg in a 10:1 ratio (pcDNA3-␤1A:pSV2*Hyg) so that transfected clones could be selected with the antibiotic hygromycin. A number of hygromycin-resistant colonies were analyzed by Northern blot for ␤1A mRNA expression. Positive clones were expanded and analyzed further by [ 3 H]STX binding. Western blot analysis of one of these cell lines, SNaIIA␤1A-16, is shown in Fig. 4, in which an immunoreactive band of approximately 45 kDa was observed.
[ 3 H]STX binding analysis revealed a significant increase in the expression levels of functional sodium channels at the plasma membrane of SNaIIA␤1A-16 cells as compared with the parent line, SNaIIA (Table I). Nonlinear regression analysis of saturation binding showed a 4.4-fold increase in B max as compared with SNaIIA with no significant change in the K D (0.8 nM for SNaIIA versus 0.9 nM for SNaIIA␤1A-16). Our results are similar to those of a previous study showing that coexpression of ␣IIA and ␤1 resulted in a 2-4-fold increase in the level of [ 3 H]STX binding compared with cells expressing ␣IIA alone (10). The K D values obtained in the present study were very similar to those reported values as well. Our data suggest that a function common to ␤1 and ␤1A is to increase the level of sodium channel expression at the plasma membrane. We hypothesize that ␤1 and ␤1A may stabilize the conformation of channels such that they become more resistant to degradation and/or target newly synthesized channels to the plasma membrane from intracellular stores. Because ␤1 and ␤1A contain a common cell adhesion molecule domain, we propose that the extracellular Ig loop may be necessary for this function.
To determine whether coexpression of ␤1A subunits affected the functional properties of type IIA sodium channels in CHL cells, we used whole cell electrophysiological recording to compare sodium currents in the parent SNaIIA cell line and in three different SNaIIA␤1A cell lines. Fig. 7A shows currents, elicited by depolarizations to varying test potentials, recorded in a typical SNaIIA cell and a typical SNaIIA␤1A cell. As is evident from these traces, coexpression of ␤1A did not dramatically alter the properties of voltage-activated sodium currents. Nevertheless, currents in ␤1A-expressing cell lines were subtly different from currents in SNaIIA cells. For example, mean steady state inactivation curves for SNaIIA␤1A cell lines were shifted to more positive potentials than the mean inactivation curve for SNaIIA cells (Fig. 7, B and D). Although this difference was quite small, it was observed in all three SNaIIA␤1A cell lines and was statistically significant in two of the three ␤1A-containing lines (SNaIIA␤1A-7, p ϭ 0.037; SNaIIA␤1A-8, p ϭ 0.024). Thus, one effect of ␤1A association with ␣IIA may be a small positive shift in the voltage dependence of steady state inactivation. For activation, mean voltage-conductance curves for two of the three ␤1A-expressing lines were statistically indistinguishable from SNaIIA (Fig. 7, B and D); however, for SNaIIA␤1A-16, the voltage dependence of activation was significantly more negative than for SNaIIA (p ϭ 0.001). Thus, data from one of the three ␤1A cell lines suggest that ␤1A may also alter sodium channel activation. For comparison, Fig. 7D also shows the effects of the adult ␤1 isoform on the voltage dependence of sodium channel activation and inactivation. As has been shown previously (10), when coexpressed with ␣IIA in CHL cells, ␤1 shifted the voltage dependence of steady state inactivation approximately 10 mV negative compared with ␣IIA alone (Fig. 7D). Thus, ␤1 and ␤1A have clearly different effects on steady state inactivation. SNaIIA␤1 cells also exhibited a small negative shift in the voltage dependence of activation compared with SNaIIA cells, as was previously reported (10).
To determine whether ␤1A affected the time course of macroscopic sodium currents, we fit the decaying phase of whole cell currents, elicited over a range of test potentials, with single exponential functions (Fig. 7C). For SNaIIA, the inactivation time constants determined from these fits were progressively less at progressively more positive test potentials, approaching a minimum of approximately 0.5 ms at the most positive test potentials examined. Inactivation time constants were significantly less for SNaIIA␤1, suggesting that ␤1 may slightly speed inactivation time course (Fig. 7C). For two of the three SNaIIA␤1A cell lines, the rate of current inactivation was virtually identical to SNaIIA (Fig. 7C). However, for SNaIIA␤1A-16, inactivation was similar to SNaIIA␤1 at all test potentials examined (Fig. 7C). Because sodium channel inactivation is coupled to activation (19,28), it is likely that the faster decay rates for SNaIIA␤1 and SNaIIA␤1A-16 currents were, at least in part, secondary to the negative shifts in FIG. 4. Western blot analysis of ␤1A expression. Brain, spinal cord, heart, skeletal muscle, adrenal, and SNaIIA␤1A-16 cell line membranes were prepared as described previously (10). The total protein in each membrane preparation was quantitated with the BCA protein assay kit using bovine serum albumin as the standard. 250 g of each membrane preparation were separated by SDS-polyacrylamide gel electrophoresis as described previously (10), transferred to nitrocellulose, and stained with Ponceau-S prior to immunodetection. Western blot analysis was performed as follows. The blot was washed for 10 min in TBS-T (10 mM Tris, pH 7.4, 150 mM NaCl, 0.1% Tween 20) at room temperature to remove acetic acid from the Ponceau-S stain and then blocked for 1 h in 5% nonfat dry milk in TBS-T at room temperature. Primary ␤1A antibody (1:750 dilution) was applied in blocking solution for 30 min at room temperature. The blot was then washed five times for 15 min each in TBS-T. Secondary antibody (horse radish peroxidase-conjugated goat anti-rabbit IgG, ICN) diluted to 1:100,000 in blocking solution was applied for 30 min at room temperature. The blot was then washed five times for 15 min each in TBS-T. SuperSignal WestFemto chemiluminescent substrate solution was applied according to the manufacturer's instructions, and the blot was placed between plastic sheet protectors and exposed to Hyperfilm-ECL for 10 s at room temperature. B, adult brain; H, adult heart; SM, adult skeletal muscle; A, adult adrenal; SC, adult spinal cord; TC, transfected cells (SNaIIA␤1A-16).
activation that were observed in these two cell lines. In addition, it is also possible that ␤1 and ␤1A subunits have direct effects on the rate of sodium channel inactivation.
The most dramatic effect of ␤1A, detected electrophysiologically, was a large increase in the amplitudes of macroscopic sodium currents. This is illustrated in Fig. 8A, which shows current densities for depolarizations to ϩ10 mV in SNaIIA, SNaIIA␤1A, and SNaIIA␤1 cell lines. Current densities for SNaIIA cells were 69 pA/picofarads, whereas current densities were approximately 2.5 times greater for the three SNaIIA␤1A cell lines (Fig. 8A). The increase in current density observed with coexpression of ␤1A was similar to that seen with coexpression of the adult ␤1 isoform (Fig. 8A). This increase in mean current density reflected two distinct effects of ␤1A on sodium channel expression. First, ␤1A greatly increased the proportion of cells with measurable whole cell sodium currents. This effect is shown in Fig. 8B, which plots the number of SNaIIA (black bars) or SNaIIA␤1A (white bars) cells with peak currents within different amplitude ranges. For SNaIIA, this amplitude-frequency distribution was bimodal. In 40% of the cells (16 of 40), currents were indistinguishable from the small inward currents recorded in untransfected CHL cells (i.e. Ͻ100 pA). In the remaining 60% of the cells, currents ranged from 500 pA to 5 nA and thus were clearly due to expression of cloned type IIA channels. In contrast, all SNaIIA␤1A cells expressed large sodium currents (Fig. 8A). The frequency his-togram for SNaIIA␤1A followed a normal distribution with a modal current range of 2-3 nA.
To determine whether the lower mean current density of SNaIIA cells was solely due to its large proportion of low expressing cells, we recalculated the mean current density for SNaIIA, excluding these low expressers. Eliminating low expressing cells increased the mean current density for SNaIIA cells from 69 to 116 pA/picofarads (Fig. 8A); however, this value was still significantly smaller than the mean current density of cells expressing ␤1A (p ϭ 0.014). Thus, even when comparing only those cells that expressed measurable sodium currents, ␤1A still increased the density of functional sodium channels on the cell surface. DISCUSSION A number of cases of intron retention have been reported in the literature, including alternative splicing of the genes encoding leukocyte-common antigen-related protein tyrosine phosphatase, CD44, effector cell protease receptor-1, the microtubule-associated protein tau, thyrotropin-releasing hormone receptor, and bovine growth hormone (29 -35). In many cases, the retained intron contains an alternate, in-frame termination codon as well as a polyadenylation signal. Alternative splicing that results in retention of the intron in the primary transcript thus results in an isoform of the protein containing a novel carboxyl terminus. Interestingly, this is not the first report of intron retention in the ␤1 gene. Waxman and co-workers (36,37) have previously reported that intron 5 of ␤1 can be retained, creating a novel isoform that is expressed in rat brain, optic nerve, sciatic nerve, and skeletal muscle. This isoform contains an 86-nucleotide insert encoded by intron 5 in the 3Ј-untranslated region. While the intron retention reported previously does not alter the ␤1 coding sequence, our present data describe a very significant coding sequence change resulting in a novel carboxyl terminus.
Effects of ␤1A on sodium currents and [ 3 H]STX Binding Levels in CHL Cells-The most striking functional consequence of ␣IIA and ␤1A coexpression in CHL cells was a 2.5-fold increase in sodium current density compared with cells expressing ␣IIA subunits alone. This increase in current density reflected two distinct effects of ␤1A: 1) an increase in the proportion of cells expressing detectable sodium currents and 2) an increase in the level of functional sodium channels in expressing cells. Increases in sodium channel expression with ␤1A are similar to previous results obtained with the adult ␤1 isoform in both mammalian cells (10) and Xenopus oocytes (4). These observations are consistent with the hypothesis that ␤1 and ␤1A subunits facilitate the expression of sodium channels and/or stabilize the channels in the plasma membrane and that the molecular basis for this function resides in the extracellular cell adhesion molecule domain common to the two isoforms. The results of our [ 3 H]STX binding experiments support this hypothesis. We propose that interaction of the extracellular cell adhesion molecule domain (Ig fold) common to ␤1 and ␤1A with ␣ may be responsible for the observed effects on channel expression levels. Consistent with this interpretation, two previous studies have shown that the extracellular domain of ␤1 is essential for modulation of both brain and skeletal muscle ␣ subunits, whereas the intracellular carboxyl-terminal domain is not; truncated ␤1 subunits lacking the intracellular carboxyl terminus are fully functional in terms of kinetic modulation of brain and skeletal muscle ␣ subunits expressed in Xenopus oocytes (13,14). Residues predicted to be in the Ig fold of ␤1 interact with type IIA ␣ subunits (14). Thus, the extracellular cell adhesion domain common to ␤1 and ␤1A appears to be required for function.
Sodium currents in ␤1A-expressing cell lines also exhibited subtle functional differences compared with the parent SNaIIA cell line. For example, inactivation curves in SNaIIA␤1A cell lines were shifted to slightly more positive potentials than inactivation curves for SNaIIA cells. In contrast as shown both here and in a previous study (10), coexpression of ␤1 with ␣IIA in CHL cells shifted inactivation to potentials approximately 10 mV more negative than for cells expressing ␣IIA alone. Perhaps the differences between these two ␤ subunit isoforms located in the putative juxtamembrane and/or transmembrane domains are responsible for these subtle distinctions in functional effects. Makita et al. (38) reported that a ␤1/␤2 subunit chimeric construct containing the extracellular region plus the first 6 residues of the transmembrane domain of ␤1 was sufficient to modulate sodium channel skeletal muscle ␣ subunits expressed in oocytes. Interestingly, this construct included an additional segment of ␤1 (ANRDMASIVSEIMMYVL) that is located carboxyl-terminal to the intron 3 splice site and is therefore not present in ␤1A. In contrast, ␤1A contains a novel juxtamembrane region that is 55 amino acids larger than that found in ␤1. This structural difference may be responsible for the opposite effects on steady state inactivation by ␤1 versus ␤1A.  In addition to opposite effects on steady state inactivation, the voltage dependence of activation and the rate of channel inactivation were also different in one of the three ␤1A-expressing cell lines, compared with the parent SNaIIA cell line. Thus, whole cell electrophysiological data suggest that ␤1A subunits may subtly modulate various aspects of sodium channel function. Nevertheless, the differences between cell lines with and without ␤1A subunits in the properties of whole cell sodium currents were very small and/or not observed in all cell lines. Therefore, additional analysis, perhaps at the single channel level, will be necessary to resolve whether these small differences actually reflect modulation of sodium channel function by ␤1A or some other source of variability between cell lines.
TTX-sensitive sodium channel ␣ subunits expressed in brain (SCN1A, Ref. 39 [42][43][44] are much less sensitive or insensitive to modulation by ␤1 when co-expressed either in Xenopus oocytes or mammalian cells. A transcript encoding a predicted TTX-insensitive sodium chan-nel (NaN/SNS-2/SCN11A) has been cloned but not yet expressed (45). Interestingly, heart, skeletal muscle, and dorsal root ganglia express both TTX-sensitive and -resistant sodium channel ␣ subunits. (Primary cultures of cardiac myocytes express SCN1A, as assessed by Western blot analysis.) 2 We have shown in the present study that these tissues also express ␤1A subunits. It will be interesting in the future to determine whether the spinal cord and brain contain unique populations of neurons expressing TTX-resistant channels and whether TTX-resistant channels are modulated by ␤1A. We also show that ␤1A subunits appear to be expressed in nonexcitable tissues such as the lung, where sodium channel ␣ subunits would not be expected to be expressed. This finding raises the intriguing hypothesis that auxiliary ␤ subunits may be expressed independently of ␣ subunits to function as cell adhesion molecules. Recent results form our laboratory have shown that ␤1 and ␤2 interact homophilically in the absence of ␣ subunits, resulting in cell aggregation and recruitment of the ankyrin cytoskeleton to the plasma membrane. 3 It will be interesting in the future to test the function of ␤1A as a cell adhesion molecule. FIG. 8. Effect of ␤1A on the level of expression of functional sodium channels. A, current densities for SNaIIA, SNaIIA␤1A, and SNaIIA␤1 cell lines. Currents were elicited by depolarization to ϩ10 mV from a prepulse potential of Ϫ100 mV. Peak current amplitude was divided by cell capacitance to give current density. Cell capacitance was determined by integrating the area under transients elicited by 3-mV voltage steps applied before series resistance compensation and capacitive transient cancellation. Mean capacitance measurements for the four different cell lines were not significantly different, indicating that coexpression of ␤1A or ␤1 did not alter cell surface area. B, amplitudefrequency histogram for SNaIIA (black bars) and SNaIIA␤1A (white bars; data for all three SNaIIA␤1A cell lines were combined). Currents were evoked by depolarization to ϩ10 mV. The bars indicate the number of cells with peak currents that fell within different amplitude ranges.