Primary structure and differential expression during development and pregnancy of a novel voltage-gated sodium channel in the mouse.

Until recently, all cloned vertebrate voltage-dependent sodium channels exhibited high sequence homology to one another and appeared to comprise a single multigene subfamily. An exception is the human Nav2.1 channel proposed to represent a second Na+ channel (NaCh) gene subfamily since comparison with previously cloned voltage-gated NaChs revealed only 40-45% identity. We have now cloned a mouse NaCh (mNav2.3) from an atrial tumor cell line that shows high amino acid sequence identity to hNav2.1 in functionally relevant regions such as the pore-forming segments, S4 segments, and inactivation gate sequence. Overall sequence identity is 68%. mNav2.3 mRNA was most abundant in heart and uterus, and the transcript levels in heart, brain, and skeletal muscle were differentially regulated during development. Transcript levels in heart were greatest immediately after birth. mNav2.3 transcript levels in pregnant uterus increased 3-fold between day 15 of pregnancy and birth and then declined 15-fold during the 2 days following delivery. The mNav2.3 amino acid sequence indicates that the Nav2 NaCh gene subfamily is well conserved across species, and the tissue-specific and developmental regulation of mRNA expression suggests these channels play important physiological roles in cardiac and uterine muscle.

B Contributed equally to this work. predict proteins that exhibit striking similarity to one another (260% overall amino acid sequence identity) and appear to comprise a single multi-gene family. This mammalian NaCh gene family has related members in such diverse species as eel (13), Drosophila (14,15), jellyfish (16), and squid (17), with the eel NaCh being most similar to the mammalian proteins. The mammalian NaCh cDNAclone that is least related to the group of similar mammalian channels is the hN%2.1 protein cloned from human heart (9). This protein shows less than 50% overall identity with other NaChs and contains unique amino acid sequence in functionally important regions that are well conserved among all other vertebrate NaChs. Hence, the hNa"2.1 NaCh has been placed in a second NaCh gene subfamily designated N%2, while all previously described NaChs can be placed in the Na,l subfamily. The hNa"2.1 channel was the first cloned NaCh to suggest multiple NaCh gene subfamilies exist and that NaCh diversity may approach that seen with K+ channels.
While the hNa"2.1 protein offers a wealth of structure function information due to its unusual amino acid sequence in functionally important regions, it has not been functionally expressed despite extensive efforts utilizing the Xenopus oocyte, Chinese hamster ovary cell, and HEK 293 cell heterologous expression systems.2 This lack of functional expression has called into question the physiological relevance of this single human cDNA clone. While expression failure could be explained by the requirement for additional subunits or specialized cellular biosynthetic machinery, other explanations include cloning artifacts or that the hNa"2.1 clone represents an expressed pseudo-gene. However, the identification of a putative Na,2 subfamily member in the rat strengthens the argument that this unusual channel is physiologically relevant. A partial cDNA sequence encoding the C-terminal one quarter of a rat glial NaCh has been reported that shows 78% sequence identity with the human Na"2.1 NaCh and less than 50% identity with other vertebrate NaChs (181, suggesting the Na,2 gene subfamily exists in nonhuman species. Since this partial clone represents the second reported member of the Na,2 subfamily, it is classified best as rat Na"2.2. However, since it was originally referred to as a glial NaCh, this designation will be used in this paper. We report here the cloning of a full-length cDNA from the mouse AT-1 atrial tumor cell line (19)(20)(21) that encodes a mammalian NaCh isoform representing the first member of the Na,2 family in the mouse. This NaCh has been designated mNa"2.3, in keeping with our earlier nomenclature (9). mNq2.3 is overall 68% identical to hNa"2.1 with similar sequence in functionally relevant regions such as pore-forming T. J. Knittle, M. M. Tamkun, and P. B. Bennett, unpublished results.

30125
Cloning and mRNA Expression of a Murine NaCh segments, positively charged S4 segments, and inactivation gate sequence. Analysis of the mNa"2.3 mRNA expression during cardiac and skeletal muscle development indicates tight developmental control over gene expression while transcript levels in uterus during pregnancy vary and dramatically decrease following delivery. Taken together, these data indicate that the Na,2 subfamily is conserved across species and likely to play important physiological roles in cardiac and uterine muscle.
EXPERIMENTAL PROCEDURES Material~-[cr~~P]dATP (3000 Ci/mmol) was purchased from ICN (Irvine, CAI. PC-1 medium was obtained from Hycor Biomedical Inc. (Portland, ME). Fetal bovine serum, penicillin, streptomycin, and the 5'-RACE system for rapid amplification of cDNA ends were from Life Technologies, Inc. Collagenase, trypsin, and dexamethasone were from Sigma. The cDNA synthesis system and oligo(dT)-cellulose, type 7, were obtained from Pharmacia Biotech Inc. The Gigapack" I1 Gold packaging extract was from Stratagene. The Sequenase version 2.0 DNA sequencing kit was purchased from U. S. Biochemical Corp. Restriction enzymes and buffers were from Boehringer Mannheim. Klenow fragment, T4 DNA ligase, and Taq polymerase were purchased from Promega. All other chemicals and reagents were of analytic grade.
Animals-DBN2J male and C57BU6J female mice (Jackson Laboratories) and their F, were used. They were mated at 60-65 days of age. . Cells were subcutaneously intaneous tumors appeared obvious, and the cells were isolated 12-16 weeks after inoculation following a method adapted from previously described methods (19,21). Briefly, tumor-bearing mice were anesthetized and sacrificed, and the tumor mass was excised, rinsed with phosphate-buffered saline (0.9% NaC1, 10 m~ NaH,PO,, pH 7.0), finely minced, and placed for 1 h a t 37 "C with gentle rocking in phosphatebuffered saline containing 100 unitdml penicillin, 100 pg/ml streptomycin, and 0.1% collagenase. The cell suspension obtained was sedimented, washed with phosphate-buffered saline, resuspended in the PC-1 medium, supplemented with penicillin (100 unitdml), streptomycin (100 pg/ml), dexamethasone (10 nM), and fetal bovine serum (lo%), and plated a t a density of 250-325 x lo3 celldml. The medium was changed every other day, and the culture grew to confluence and was usually beating spontaneously within 10-14 days in culture. Total RNA Extraction, Poly(A)+ RNA Purification, and Northern Blot Analysis-Total RNA was isolated using the guanidinium thiocyanate method, and poly(A)' RNA was purified through oligo(dt)-cellulose chromatography by standard procedures (9,22). For Northern blot studies, up to 20 pg of total RNA were fractionated by electrophoresis through a 1% agarose, 3% formaldehyde gel in 20 mM MOPS, 1 m~ EDTA, pH 7.4. Application of equal amounts of RNA to each lane was confirmed by the addition of ethidium bromide to the samples prior to electrophoresis and the quantitation of the stained ribosomal bands by means of a imaging densitometer (Bio-Rad). Less than 15% variability was observed among lanes. The gel was submerged for 5 min in 50 mM NaOH and 1.5 M NaCl and then neutralized 30 min in 1 M Tris, pH 6.8, and 1.5 M NaCl before a n overnight transfer to a Nytran filter (Schleicher and Schuell) by capillary action in 20 x SSC (3 M NaC1, 300 mM sodium citrate, pH 7.0). RNA was cross-linked to the filter by irradiation with ultraviolet light for 3 min. The filters were prehybridized overnight a t 65 "C in 20% formamide, 10% dextran sulfate, 4 x SSPE (600 mM NaCI, 40 mM NaH,PO,, 4 mM EDTA, pH 7.4), 5 x BFP (1 gfliter bovine serum albumin, 1 glliter polyvinylpyrrolidone 40, 1 ghiter Ficoll, 0.001% SOdium azide), 0.1 mg/ml sonicated salmon sperm DNA, 0.2 mg/ml yeast RNA, and 5% SDS. Filters were hybridized 24 h at 65 "C in the same solution with lo6 c p d m l of an [O~-~~P]ATP random primer-labeled cDNA probe (clone C291, corresponding to the isoform-specific C-terminal and 3'-untranslated region). Filters were washed for 30 min at 65 "C with 3 x SSC and 1% SDS, 30 min with 1 x SSC and 1% SDS, and 30 min with 0.2 x SSC and 1% SDS before autoradiography. The quantitation of the specific hybridization was directly assessed by means of a PhosphorImager (Molecular Dynamics). At least three different filters made from pooled samples were analyzed in each experiment. Every pool had RNA from at least 5, 60, and 20-60 adult, fetal, and newborn animals, respectively. Thus, 5-7 different mothers and their respective offspring have been analyzed in every point. Values were subtracted from the background, corrected, and standardized by hybridization to the 28 S ribosomal band.
cDNA Library Construction, Screening, and Sequence Analysis-A A g t l O library was constructed from AT-1 cells using a cDNA synthesis system as previously described (9). Size-fractionated cDNA (>2 kilobases) was inserted into A g t l O arms using EcoRI-Not1 adapters. Up to 5 x lo5 primary plaques were screened at low stringency with a cDNA containing the hNq2.1 coding sequence as previously described (9). One positive recombinant (clone C291) was isolated, amplified, and subcloned into the EcoRI site from pGEM7 (Promega). Using clone C291 as a probe, 4 x lo6 primary plaques were screened a t high stringency.
Positive recombinants were isolated, amplified, and subcloned into the Not1 site from pBluescript I1 KS(+) (Stratagene). DNA sequence analysis of both strands was performed as previously described (9).
Rapid Amplification of cDNA Ends and PCR Amplification-A 5'-RACE reaction was performed as described in the manufacturer's instructions with several modifications. Briefly, cDNA was synthesized from 1 pg of poly(A)' RNA from the cultured AT-1 cells using the 5'-ACCAAATM'CCAACGTGAAGGAGA-3' gene-specific primer (GSP-1) corresponding to the 5'-end of the C171 clone. Superscript I1 reverse transcriptase was used. The cDNA was dC-tailed using dCTP during 30 min at 37 "C. Using the nested primer 5"AGGAGAGACCTGAGCTC-CTCGCTC-3' (GSP-2) and the manufacturer's anchor primer (5'-(CUA),GGCCACGCGTCGACTAGTACGGGIIGGGIIGGGIIG-3'), the specific cDNA was amplified using Taq polymerase. An aliquot of this reaction was taken and used for a second PCR amplification. To increase the specificity and yield, a new gene-specific primer (GSP-3, 5'-(CAU),n"rCCAGCTGAAGCAGGT-3') was used that targeted sequence about 200 base pairs upstream in the C171 clone. The primers used in the second reaction were the GSP-3 and the manufacturer's Universal Amplification Primer (5'-(CUA),GGCCACGCGTCGACTAGTAC-3').
The positive product was isolated, characterized, blunted with Klenow fragment, and subcloned into the SmaI site of the pGEM 7. This clone (RC715) was sequenced and found not to contain the complete coding sequence. The 5'-RACE was repeated starting with cultured AT-1 cell poly(A)' RNA to obtain the full coding sequence. The new primers used were GSP-I,

RESULTS AND DISCUSSION
Cloning of the mNaB.3 Channel from Mouse AT-1 Cells-A mouse atrial tumor cell line (AT-1) was chosen for this work since the study of ion channels is simplified in cell lines where individual cells can be studied by voltage clamp and populations studied with cellular and molecular techniques. AT-1 cells, derived from atrial tumors in transgenic mice carrying fusions between the atrial natriuretic factor promoter and the SV40 large T antigen coding sequence, retain a highly differentiated cardiac phenotype in culture, including expression of adult cardiac-specific proteins, spontaneous beating, and secretion of atrial natriuretic factor (19)(20)(21). A cDNA library was constructed from a beating AT-1 cell preparation and screened at low stringency with hNa"2.l-derived probes as described under "Experimental Procedures." Four cDNA clones were isolated and sequenced as shown in Fig. 1. The longest clone (C171) was 5 kilobases in length and contained 75% of the coding sequence. Repeated screening of the library with 5'fragments of the C171 clone failed to isolate additional sequence, necessitating the use of 5"RACE-PCR. Two successive rounds of RACE produced the two fragments listed as RC715 . 1. c D N h encoding the m N 0 . 3 clone. The bur represents the longest open reading frame encoded by these overlapping cDNAs, and untranslated regions are shown by the flanking black lines. Clones C291, (2151, C171, and C191 were isolated from the AT-1 cDNA libraw.

I -C151
Clones RC715 and RB1401 were obtained by 5'-RACE using AT-1 cell first strand cDNA as described under "Experimental Procedures." and RB1401 in Fig. 1. All cDNA fragments were sequenced on both strands, and the sequence of PCR products from separate amplifications was compared to guard against polymeraseinduced sequence error.
The complete sequence, designated mNav2.3, consists of 251 base pairs of 5"untranslated sequence, an open reading frame of 5043 base pairs, and at least a 2.0-kilobase 3"untranslated region. A polyadenylation signal sequence (AATAAA) and a poly(dA) tail were not found. The nucleotide sequence immediately surrounding the assigned initiation codon ( C G M L G T , nucleotides 247-254) resembles the consensus eukaryotic initiation sequence (23) only in the presence of a cytosine nucleotide at position -5 and an adenosine nucleotide at position -3 (relative to the start codon). Two in-frame termination codons ( T U and TGA) are present starting a t nucleotide 5295. Interestingly, 5' of the assigned start codon in previously cloned vertebrate NaChs belonging to the NaJ family is an out of frame ATG triplet a t position -8 relative to a typical eukaryotic initiation sequence ( m C C A A C a G ) . However, in the human Na"2.1 sequence, while this motif was present 5' to the assigned initiation codon, both ATGs preceded short open reading frames. The mNq2.3 channel lacks the upstream ATGs, and the first in frame methionine codon corresponds to the predicted translation start site in the human Na"2.1 channel. No other potential translation start sites exist within the 5'untranslated region.
The deduced amino acid sequence of the AT-1-derived sequence is shown in Fig. 2 and compared with the full-length human Nq2.1 (9), H1 from rat heart (€0, and the rat glial partial sequence (18). The deduced primary structure of mNa"2.3 consists of 1681 amino acids and has a calculated molecular weight of 192,144. Prediction of transmembrane topology by hydropathy analysis reveals a profile similar to other NaChs, with four large (229-280 residues) hydrophobic domains, each composed of a t least six potential membrane spanning a-helical segments including a positively charged amphipathic segment (S4). As with the original hNq2.1, mNa"2.3 shares no significant overall amino acid identity with voltagedependent potassium channels or other known protein sequences. The amino acid identity with voltage-dependent Ca2+ channels is less than 10% (24). Comparison of mNq2.3 with each of the five complete rat NaCh sequences reveals a uniform pattern of overall primary structure homology (overall amino acid identity is 39% for rat brain I (4), 34% for rat brain I1 (4), 33% for rat brain I11 (5),33% for rat skeletal muscle SkMl(11), and 33% for rat heart I (8)). However, mNa"2.3 is more homologous to hNq2.1(68% amino acid sequence identity) than it is to these rat channels, indicating a high degree of relatedness to N q 2 subfamily of NaChs. Significant homology is also evident from comparisons of mNa"2.3 with NaChs from the eel electric organ (31% amino acid identity) (13), the Drosophila para locus (22%) (151, and squid (22%) (17). Only 15% identity was ob- Regional comparisons of mNa.9.3 with other NaChs reveal a high degree of homology within the repeat domains where up to 84% amino acid identity is observed (Table I). In contrast, there is poor conservation of primary sequence within the interdomain regions ID1-2 and 1D2-3. Clearly, the mNq2.3 channel is most similar to the hNa"2.1 clone than other mammalian channels. A high degree of amino acid sequence identity (-74% when comparing mNq2.3 and rat heart I) with other NaChs is found within two short segments (SS1 and SS2) of the S5-S6 interhelical region (Fig. 2) that are believed to form membrane-penetrating hairpin structures that contribute to the formation of the ion pore (25) and various neurotoxin binding sites (26,27).
The primary sequence of mNa"2.3 shares distinct features with hNa"2.1 in regions known to be important in voltagedependent activation and inactivation. The S4 segments of these two channels, which may function as voltage-sensors, collectively exhibit fewer positive charges than is typical of other NaChs, vertebrate or invertebrate (Fig. 2). The greatest differences occur in domain 4 where there are only 4 arginine or lysine residues as compared with 8 for other NaChs. Histidines (residues 1355 and 1367) replace arginines at two positions in the 54 segment of domain 4. Other S4 segment variations include the substitution of glutamines (residues 208 and 1345) for arginines in domains 1 and 4, and aliphatic residues (Ile-1029, Leu-1349) for arginines in domains 3 and 4. The S4 segment of domain 2 in mNq2.3 has a reduced number of positively charged residues (2 instead of 5 , with three arginines being replaced with leucine, valine, and isoleucine). This reduction in positive charge in the S4 segment of domain 2 represents the greatest difference between the mNa"2.3 and hNr~2.1 channels within functionally significant regions. All other NaChs, with the exception of the jellyfish NaCh, which lacks the fifth positive residue, contain five positive amino acids in this region. The ID3-4 region, which has an essential role in inactivation (28), is unlike the sequence found in most NaChs. The replacement of the highly conserved protein kinase C phosphorylation site (found in all previously described NaChs except jellyfish and hNq2.1) with a potential tyrosine kinase site (29) is notable. Within the putative pore-forming region, the mNq2.3 channel has amino acid substitutions at 2 residues involved in determining ion selectivity; position 1142 is an Asn as opposed to the Lys found in all other NaChs except the hNq2.1 and jellyfish NaChs, where this position is also a n h n , and position 1434 is a Ser as opposed to the Ala found in all other NaChs except the rat glial channel.
There are five potential sites for N-linked glycosylation in regions of hNq2.1 predicted to be extracellular, while in mNq2.3 there are six sites. Like other NaChs, these potential sites are clustered in the S5-S6 interhelical regions of domains 1 and 3. Potential sites for cyclic nucleotide-dependent phos- phorylation are also well conserved between the two channels, heing present in the ID1-2 fSer-443) and ID24 (Ser-756, Ser-857, Ser-869, and Srr-906) regions of mNa,.2.3. A single sitr conserved hetwren the two proteins in the C terminus also exists at Ser-1511. The drnsitv of these sitrs in the ID1-2 region is much less than in rat hrain NaChs (1 ).
in RNA from hrain, kidney, and skeletal muscle, hut no specific hyhridization was ohserved in RNA from liver. These results are consistent with the tissue-sprcific expression of mNa,2.3 primarily in heart, skeletal muscle, and uterus. This tissue specificity of expression was similar to that descrihrd for t h r human Na,.2.1 channel (9). Voltage-gated ion channel gene expression varies during development presumahly to satisfv tissue-specific needs, and strict developmental regulation may he taken as a n indication of the physiological importance of a given channel to tissue function. Therefore, the time course of expression of mNa,2.3 during development was determined in mouse heart, skeletal muscle, and hrain (Fig. 4). mNq2.3 expression increased approximately 2.5-fold in emhryonic heart just prior to hirth and then decreased at least 20-fold hy day 7 after hirth. Expression levels then slowly increased to the adult level, which was half that ohserved just prior to hirth. In skeletal muscle, levels slowly increased throughout development reaching a peak at 21 days oflife, which was then followed by 6-fold decline to adult levels. Expression in hrain showed a pattern similar to heart in that expression peaked just prior to birth and then declined approximately IO-fold by day T of life. An increase of severalfold was ohserved between days 7 and 14, and after day 21 the rxpression level increased at least 6-fold to the adult levels. Thus, the mNa,.2.3 transcript is well regulated during emhryonic and neonatal development in a tissue-specific manner.
Developmental regulation of hoth NaCh mRNA and function has heen extensively studied, especially in skeletal muscle. Early in muscle development, the predominant NaCh is a TTXresistant type represented hy the heart type I isoform (8,12,30). Beginning several days after hirth, the resistant channels are replaced by the TTX-sensitive isoform characteristic of adult skeletal muscle, and this exchange is complete within 30 days. Changes in mRNA levels parallel the developmental changes in Na+ currents (12,30). Muscle NaCh transcript levels vertebrate NaCh gene subfamily 1 (Na,l). The hNa"2.1, mNa"2.3, and rat glial NaCh are grouped in subfamily 2 (Na,2), most likely representing a second, evolutionarily distinct subfamily for vertebrates. The invertebrate channels from squid and Drosophila are most similar to each other, while the jellyfish channel stands alone as expected since the jellyfish represents the evolutionarily oldest animal with a nervous system.
In comparing ion channel amino acid sequence differences across species, it is often difficult to distinguish between isoform differences and simple species variation. The mNa"2.3 sequence is 68% identical to that of the human N%2.1 clone, with the sequence differences evenly dispersed throughout the protein as opposed to being concentrated in nonmembranespanning regions, which are likely to be less important for channel function. Cross-species amino acid sequence identity between the same NaCh isoform is typically greater than 90%. For example, the skeletal muscle I(10, 111, brain type I1 (4, 61, and cardiac I (7,8) NaChs share 92, 97, and 90% identity between human and rat species, respectively. Therefore, it is tempting to speculate that the mNa"2.3 clone represents a related but distinct NaCh isoform when compared with the hNa"2.1 NaCh. The difference in the number of unusually conserved positive charges in the second S4 region between the human and mouse clones does argue for two distinct isoforms. However, the similarity in the tissue specificity of transcript expression between human and mouse suggests that these two clones represent the same isoform from a functional standpoint.
While only the C-terminal quarter of the glial NaCh amino acid sequence has been reported (18), this sequence is 90% identical with the corresponding region of Nq2.3, raising the possibility that these two clones represent species homologs. In addition, the patterns of mRNA expression are similar. The glial channel was expressed in heart, and only low levels were observed in brain. Expression in uterus was not determined. However, as shown in Fig. 4, mNq2.3 mRNA was high in mouse heart at PN1 while the rat glial NaCh mRNAwas low in rat heart at PN1 relative to adult levels (18). Without the full-length rat glial sequence, it is not possible to determine whether the mNa"2.3 sequence reported here represents the same isoform as the rat glial protein, for C-terminal similarity may be misleading. The rat brain I1 and I11 NaChs represent different gene products, are 85% identical overall, but show 89% identity in repeat IV and the C terminus. It is premature to draw any conclusions with respect to the relatedness of the human Na"2.1, mouse Nq2.3, and the glial channel beyond the idea that they are clearly within the same NaCh gene subfamily. CONCLUSION The mNa"2.3 cDNAclone provides the first full-length amino acid sequence for a putative rodent NaCh that has unique features in regions known to be important in voltage-dependent activation and inactivation. mNa"2.3 is homologous to the human Na"2.1 clone and can be placed in the Na,,2 NaCh gene subfamily. The conservation of usual sequence in functionally important domains between the hNq2.1 and mNq2.3 channels and the similar tissue-specific expression between human and mouse argue in favor of an important physiological role for members of this second mammalian NaCh gene subfamily. In addition, up-regulation of transcript levels in mouse uterus near delivery suggest an important role in control of uterine excitability. The mouse species will be valuable in determining Eel J e l l y f i s h the physiological role that these Nq2 subfamily NaChs play in tissue physiology. Unanswered questions concern the specific cells expressing the mNq2.3 channel, its subunit composition, and the nature of the ionic current generated. Site-directed antibodies will be required to address the first two points while heterologous expression, if successful with the mNq2.3 clone, will provide insight into the current phenotype. If additional subunits are required for expression, the AT-1 cells and native mouse tissue will provide the means by which to biochemically identify these proteins.