|Home | About | Journals | Submit | Contact Us | Français|
The voltage-gated sodium channel Nav1.8 is only expressed in subsets of neurons in dorsal root ganglia (DRG) and trigeminal and nodose ganglia. We have isolated mouse partial length Nav1.8 cDNA clones spanning the exon 17 sequence, which have 17 nucleotide substitutions and 12 predicted amino acid differences from the published sequence. The absence of a mutually exclusive alternative exon 17 was confirmed by sequencing 4.1 kilobases of genomic DNA spanning exons 16–18 of Scn10a. A novel cDNA isoform was identified, designated Nav1.8c, which results from alternative 3′-splice site selection at a CAG/CAG motif to exclude the codon for glutamine 1031 within the interdomain cytoplasmic loop IDII/III. The ratio of Nav1.8c (CAG-skipped) to Nav1.8 (CAG-inclusive) mRNA in mouse is ~2:1 in adult DRG, trigeminal ganglion, and neonatal DRG. A Nav1.8c isoform also occurs in rat DRG, but is less common. Of the two other tetrodotoxin-resistant channels, no analogous alternative splicing of mouse Nav1.9 was detected, whereas rare alternative splicing of Nav1.5 at a CAG/CAG motif resulted in the introduction of a CAG trinucleotide. This isoform, designated Nav1.5c, is conserved in rat and encodes an additional glutamine residue that disrupts a putative CK2 phosphorylation site. In summary, novel isoforms of Nav1.8 and Nav1.5 are each generated by alternative splicing at CAG/CAG motifs, which result in the absence or presence of predicted glutamine residues within the interdomain cytoplasmic loop IDII/III. Mutations of sodium channels within this cytoplasmic loop have previously been demonstrated to alter electrophysiological properties and cause cardiac arrhythmias and epilepsy.
Voltage-gated sodium channels mediate the rapid influx of sodium ions that initiate action potentials in excitable cells. The large pore-forming α-subunits of ~260 kDa contain four internally homologous transmembrane domains (I-IV) connected by the three interdomain cytoplasmic loops IDI/II, IDII/III, and the smaller IDIII/IV (1). Nine different mammalian type 1 α-subunits have been cloned, designated Nav1.1–Nav1.9 in the unified nomenclature (2), some of which have multiple isoforms because of mutually exclusive alternative splicing. For example, Nav1.2, Nav1.3, and Nav1.7 each utilize either a neonatally expressed exon 5N or a predominantly adult-expressed exon 5A, which differ by 20 or 21 nucleotides and encode differences of only 1 or 2 amino acids (1). Homologous exons 5N and 5A have also been identified in the gene encoding Nav1.6 (3).
The Nav1.5/1.8/1.9 subfamily is resistant to the neurotoxin tetrodotoxin (1, 4), with Nav1.5 being the principal cardiac subtype (1, 2), although it is also expressed at much lower levels in brain, embryonic dorsal root ganglia (DRG),1 and embryonic skeletal muscle (5-9). Nav1.5-null mutant mice die around embryonic day 11.5 with severe defects in cardiac ventricle morphogenesis (10). Expression of Nav1.8 (formerly SNS/PN3) is restricted to small diameter sensory neurons in DRG, trigeminal ganglia, and nodose ganglia (4, 11), whereas Nav1.9 (formerly NaN/SNS2) is preferentially expressed in small diameter DRG and trigeminal ganglia neurons, with much lower levels in retina and brain (12-14). The down-regulation of Nav1.8 expression in rat DRG using antisense oligodeoxynucleotide can prevent thermal hyperalgesia (hypersensitivity to noxious stimuli) and allodynia (pain response to non-noxious stimuli) in a model of neuropathic pain, whereas down-regulation of Nav1.9 has no apparent effect on pain behavior (15, 16). Oligodeoxynucleotide knockdown of Nav1.8 also decreases inflammatory hyperalgesia induced by prostaglandin E2 (17), and suppresses nociceptive responses in a rat model of visceral pain (18). Nav1.8-null mutant mice demonstrate deficits in the responses to noxious mechanical and thermal stimuli (19), an absence of spontaneous activity in damaged sensory neurons (20), a delayed development of inflammatory hyperalgesia (19), and deficits in visceral pain (21).
The α-subunit cytoplasmic loops possess ordered secondary structures (22) and, in a yeast two-hybrid screen using rat Nav1.8 as bait, have been shown to mediate interaction with accessory proteins, e.g. the IDII/III segment is bound by papin (PDZD2/AIPC) and by a protein similar to syntrophin-associated serine/threonine kinase (23). A growing body of data indicates that the cytoplasmic IDII/III loop can modulate channel function, including channel activation and slow inactivation, and interaction with the membrane-cytoskeletal adaptor protein ankyrin-G (24-26). Single amino acid changes within the IDII/III loop can alter channel electrophysiological properties and have been implicated in disease, e.g. the Nav1.5 variant S1102Y present in 13.2% of African Americans is over-represented in patients with cardiac arrhythmias (27). In addition, the Nav1.1 variant W1204R is associated with the epilepsy syndrome GEFS+ (generalized epilepsy with febrile seizures plus, Refs. 28-30), and among Nav1.5 variants, A997S was present in a case of sudden infant death syndrome (SIDS), S941N in a near-SIDS case of long-QT syndrome (LQTS), and D1114N in LQTS (31-33).
We have isolated a novel Nav1.8 cDNA isoform, Nav1.8c, that excludes the codon for a glutamine residue within the IDII/III loop, caused by alternative 3′-splice site selection at a CAG/ CAG motif in both mouse and rat. No equivalent alternative splicing of Nav1.9 was detected. The isolation of a novel isoform of Nav1.5 cDNA that encodes an additional glutamine residue confirmed the occurrence of analogous alternative splicing at a CAG/CAG motif in both mouse and rat Nav1.5. These results have potential implications for the function of Nav1.5 in cardiac physiology and for Nav1.8 in pain perception.
All animals were fed standard chow and water ad libitum, and animal care procedures were carried out within UK Home Office protocols and guidelines.
Tissues from adult male 129/OlaHsd mice killed by cervical dislocation were frozen on dry ice and stored at −80 °C. To study changes in expression after axotomy, age-matched 10–12-week-old male mice were anesthetized, and the right sciatic nerve transected as previously described (34), prior to killing 7 days later to obtain ipsilateral (axotomized) and contralateral (control) lumbar L4 and L5 DRG pools from six animals. Neonatal (postnatal day 1, P1) DRG or hearts were pooled from seven 129/OlaHsd pups. Two adult male Wistar rats (~300 g) from the Bristol University breeding colony were killed by anesthetic overdose, and tissues were harvested.
Total RNA from DRG and neonatal heart was purified using an RNeasy Mini Kit (Qiagen), with tissue homogenization using 21-gauge and 25-gauge needles, and final elution in 60 μl of RNase-free water (Qiagen). RNA was then treated with 10 units of RQ1 RNase-free DNase (Promega) for 30 min, which was inactivated during re-extraction of RNA (35) following a RNA cleanup protocol (Qiagen). RNA concentration was determined by absorbance at 260 nm, and 1 μg used per 50 μl of reverse transcription (RT) reaction with random hexamer primers (2.5 μM; TaqMan Reverse Transcription Reagents Kit, Applied Biosystems).
Trigeminal ganglia and lumbar cord were processed with a Power-Gen 125 homogenizer and disposable Omni-Tips 7-mm diameter generator probes (Fisher Scientific), with total RNA processing as above. Larger tissues (e.g. adult heart and thigh muscle) were homogenized using a mortar and pestle, with total RNA isolation using RNAzol-B (Biogenesis) followed by DNase treatment and re-purification as above.
Four primers (Invitrogen) were designed for nested RT-PCR to amplify a portion of the Nav1.8 interdomain cytoplasmic loop IDII/III coding region ((11); GenBank™/EBI accession number Y09108, Ref. 36), and which would cross intron 17 of ~2.3 kb (36): 5′-CCCAAAGGGCAGCAGGAGCTG-3′ (SNSfor1); 5′-CGGCGAGTGCAGCCTTCTGTGA-3′ (SNSrev1); 5′-CTGCCACAAGTCCAAAAGTGTGAA-3′ (SNSfor2); and 5′-AGTCATCGGGCTCGTCCAGATC-3′ (SNSrev2). Sequences correspond, respectively, to nucleotides (nt) 3114–3134, 3403–3382 (negative strand), 3135–3158, and 3378–3357 of Y09108. Underlined primer nucleotides correspond to either the skipped CAG trinucleotide or nucleotides for which substitutions are described in this report (C3139A, A3140G).
50 μl of RT-PCR used 1 μl of RT reaction (20 ng of total RNA equivalent) with primer pair SNSfor1/SNSrev1, and either recombinant TaqDNA polymerase (Invitrogen, 10342) or a hot start DNA polymerase that includes a proofreading enzyme (Clontech Advantage cDNA polymerase, 8417-1), under the following conditions: 94 °C for 2 min, and 35 cycles of 94 °C, 30 s; 64 °C, 45 s; 72 °C, 30 s, with a final 72 °C for 10 min. The single apparent product was cut out from a 2% agarose gel, purified (Gel Extraction kit, Qiagen), and used in a 25-cycle nested PCR with primer pair SNSfor2/SNSrev2, prior to cloning products into pCRII-TOPO using a TOPO TA cloning kit and transformation into One Shot TOP10F′ competent cells (Invitrogen).
Primers to amplify larger Nav1.8 partial cDNAs transcribed from Scn10a exons 16–18 were: 5′-TGCCGGTTCCGTTGGCCCAAGGT-3′ (SNSfor3); 5′-AGCTGCAAAGTTGAGAACCACATTGC-3′ (SNSfor4); and 5′-GATCTCCTCTGGGTCCGGGCAGT-3′ (SNSrev3). These sequences correspond, respectively, to nt 2838–2860, 2895–2920, and 3329–3307 of Y09108, with a nucleotide for which a substitution is described in this report (T2905C) underlined. RT-PCR conditions using either TaqDNA polymerase were as above, except for extension times of 45 s, and the use of both 20 and 100 ng of total RNA equivalents for (negative control) neonatal heart. The purified product was used in a 25-cycle seminested PCR with primer pair SNSfor4/SNSrev3, prior to either purified PCR products (PCR purification kit, Qiagen) or DNA minipreps (Qiagen) of TA-cloned products being incubated with Bsu36I (New England Biolabs), which does not cut the vector pCRII-TOPO.
Nav1.8 cDNA inserts were also cloned from trigeminal ganglia, as above, using the primer pair SNSfor3/SNSrev3. Note that the resultant product identified the T2905C substitution within the SNSfor4 primer sequence (underlined above).
Primers to amplify exon 17 of Scn10a from genomic DNA were: 5′-GTTGATAATGACTCTAATTCCCAG-3′ (SNSfor5); 5′-AATTCCCAGACGGCCCTCTCA-3′ (SNSfor6); and 5′-ACTCTACCTCGGCAGGGACC-3′ (SNSrev5). The forward primers are from available intron 16 sequence, while the reverse primer corresponds to all available 5′ intron 17 sequence (Ref. 36, Fig. 2) plus nt 3266–3254 of Y09108. 50 μl PCRs each used 1 μl of a 50-μl 129/ OlaHsd mouse tail DNA preparation (kindly provided by Rob Pope, LINE, Bristol University) and recombinant TaqDNA polymerase under the following conditions: 94 °C for 5 min, and 35 cycles of 94 °C, 40 s; 65 °C, 1 min; 72 °C, 30 s; with a final 72 °C for 10 min prior to TA-cloning. No amplification product of the expected size resulted from primer pair SNSfor5/SNSrev5 when annealed at 65, 63, or 60 °C (data not shown). Genomic clones from PCR with SNSfor6/SNSrev5 with the expected size of 173 bp were screened for cleavage by Bsu36I, and four clones were sequenced. Our subsequent intron 16 sequencing revealed four nucleotide insertions and two deletions relative to the primer SNSfor5, and four nucleotide insertions relative to the primer SNSfor6 (AJ622906, respectively, nt 1722–1747 and 1738–1762).
Primers used to amplify intron 16 of Scn10a were: SNSfor3, SNSfor4, and 5′-CCCCCAGGTATGGAGCCAGGT-3′ (SNSrev6) corresponding to nt 3225–3205 of Y09108. PCR conditions with primer pair SNSfor3/ SNSrev6 were as for exon 17, except for extensions of 1 min 30 s, and the purified (Qiagen) product of ~1.8 kb was used in a 30-cycle seminested PCR prior to TA-cloning (Invitrogen). Primers for the amplification of intron 17 were: 5′-CACCCTCCGGGATGTCCTCTGA-3′ (SNSfor8; nt 3181–3202 of Y09108, except for the A3183C substitution described in this study); 5′-CGCCAGCTCAGGGATCTTCCTCA-3′ (SNSrev7; nt 3353–3331 of Y09108); and SNSrev3. PCR conditions were as for intron 16, with the SNSfor8/SNSrev7 product of ~2.3 kb purified prior to seminested PCR and TA-cloning. Four clones each of intron 16 and intron 17 were DNA sequenced in both directions using a panel of internal primers (details available on request).
Primers to amplify a partial length rat Nav1.8 cDNA were 5′-CTCAGAGGCCAAGAACCACATTGC-3′ (rSNSfor4) and 5′-AGCTCGGGGATCTTCCTCAGGAT-3′ (rSNSrev1). These correspond, respectively, to nt 3068–3091 and nt 3520–3498 of X92184 (SNS, Ref. 11), which are identical in U53833 (PN3, Ref. 4). RT-PCR used 20 ng of total RNA equivalent and recombinant TaqDNA polymerase under the same conditions as mouse Nav1.8 cDNA exons 16–18 (see above).
Primers were designed to amplify a portion of the IDII/III coding region of mouse Nav1.5 cDNA (mH1, Ref. 37; AJ271477) spanning DNA corresponding to human SCN5A (Nav1.5) exons 17–19 (37, 38) that are analogous to mouse Scn10a (Nav1.8) exons 16–18 (36), with primers encoding residues dissimilar to Nav1.8 (11). The primers 5′-AAACACGGTTCGAGGAAGACAAGC-3′ (5for1) and 5′-GGAGGTCTGCGGTGTTGGTCATG-3′ (5rev1) correspond, respectively, to nt 3130–3153 and 3507–3485 of AJ271477 (37). RT-PCR of either 20 or 100 ng of total RNA equivalents (1- or 5-μl RT reactions) used recombinant TaqDNA polymerase followed the same conditions as for mouse Nav1.8 cDNA exons 16–18 (see above), except for either 35 or 40 cycles. Products subsequently cloned were derived from: 1 μl of RT, 35 cycles (neonatal and adult heart); 5 μl of RT, 35 cycles (brain); or 5 μl of RT, 40 cycles (neonatal DRG and adult thigh muscle). Full-length Nav1.5 (378 bp) and smaller isoform Nav1.5a (219 bp) bands were cut out from 2.5% agarose gels and purified. Full-length Nav1.5 cDNAs were TA-cloned (as above). The Nav1.5a cDNAs were used in 25 cycle seminested PCRs with primers 5rev1 and 5for2 (5′-GCTGAGTCAGACACTGATGACCAG-3′, corresponding to nt 3210–3233 of AJ271477), and the expected products of ~139 bp were TA-cloned.
Primers to amplify Nav1.9 cDNA transcribed from Scn11a exons 14–15B were: 5′-CCCTGGATACAAGGTCCTGGAAG-3′ (NaNfor1), 5′-TGATTCAGAAATGACTCTGTACACTG-3′ (NaNfor2), and 5′-CAACTGAGGCCCTTGGGAAAGCA-3′ (NaNrev1) corresponding, respectively, to nt 2661–2683, 2689–2714, and 3063–3041 of AF118044 (39). The amplicon crosses both intron 14 of ~1.8 kb and intron 15A of 1001 bp (39), and primer sequences are identical to another Nav1.9 cDNA sequence (Ref. 40, AB031389). Nav1.9 cDNA was amplified from 20 ng of total RNA equivalent (1-μl RT reaction) from neonatal DRG, using recombinant TaqDNA polymerase for 35 cycles under the same conditions as for mouse Nav1.8 cDNA exons 16–18 (see above). Purified product was subjected to seminested PCR for 25 cycles and TA-cloned. Nav1.9 cDNA clones were tested for the presence of the single expected StuI and BbsI (New England Bio-labs) restriction enzyme recognition sites at, respectively, nt 2820 and 2839 of AF118044, such sites being absent in the vector pCRII-TOPO.
Primers were designed to amplify a rat cDNA encoding part of the IDII/III cytoplasmic loop (Ref. 41, M27902) of the full-length Nav1.5 cDNA isoform, but not of the Nav1.5a/H1–2 isoform (see above), to tend to avoid encoded residues conserved in Nav1.8 (11), to cross the sequence analogous to human SCN5A intron 17 (38), and to span the exon 18/19 junction (38). The primers 5′-CTGCGGCGGCGACCTAAGAAGC-3′ (r1.5for1), 5′-GTCCTCAGGGGTCTCACCGCAC-3′ (r1.5rev1), and 5′-CCAGTCTGCCTGAGATGTACTGG-3′ (r1.5rev2) correspond, respectively, to M27902 nt 3159–3180, 3605–3584, and 3548–3526, which share identity with the sequence of Kallen et al. (Ref. 8, SkM2). 20 ng of total RNA equivalent (1-μl RT reaction) was used in a 40-cycle PCR, under the same conditions as for mouse Nav1.8 exon 16–18 cDNA (see above), and the purified single apparent product was used in a 25-cycle seminested PCR with primer pair r1.5for1/r1.5rev2.
All DNA sequencing was performed by the Department of Biochemistry, University of Oxford, and chromatogram figures utilized Chromas version 2.23 (Technelysium). DNA data base searches used FASTA (42), and the presence and classification of repetitive elements within genomic DNA was determined using RepeatMasker (available online at repeatmasker. genome. washington. edu/cgi-bin/RepeatMasker) within NIX URL: www.hgmp.mrc.ac.uk/NIX/). Protein motif recognition used PROSITE (available online at ca.expasy.org).
Nav1.8 cDNA encoding part of the inter-domain cytoplasmic loop IDII/III was cloned from adult mouse DRG by RT-PCR, using primers based on the published mouse cDNA sequence that was derived from genomic DNA (Ref. 36, GenBank™/EBI accession no. Y09108). However, all five sequenced cDNA clones had ten identical nucleotide substitutions within the 198-bp insert, as did ten clones from subsequent RT-PCRs using two different TaqDNA polymerases (data not shown, see “Experimental Procedures”). As the substitutions were all within the 108-nt portion transcribed from exon 17 of Scn10a (36), it was possible that the clones each represented a mutually exclusive exon 17 splice variant of Nav1.8. Therefore, PCR primers were designed to amplify Nav1.8 cDNA transcribed from exons 16 to 18, such that the plasmid containing the novel sequence would not be cut by the restriction enzyme Bsu36I, unlike the published sequence that has a recognition site at nt 3249 (Y09108).
A PCR product of the expected size (491 bp) was made from reverse-transcribed RNA of neonatal, adult-control, and adult-axotomy (nerve-transected) DRG. No product was detected in RT-minus controls, or by RT-PCR of neonatal heart RNA (Fig. 1), as previously shown for rat and human heart (4, 11, 43). Consistent with the novel cDNA sequence, there was no apparent digestion by Bsu36I either of the 435-bp seminested PCR products prior to cloning, or of 8 individual cDNA clones from each of the three DRG tissues (data not shown). The cDNA clone DNA sequences each confirmed the 10 base differences described above. In the extended available sequence, all cDNA clones had a further two differences in the complete exon 17-transcribed sequence (total 12/141 nt) and four differences in the 205 nt of exon 16-transcribed sequence. Table I details the nucleotide and predicted 11 amino acid differences from the previously reported Nav1.8 sequence (36), and Fig. 2 shows an alignment of the predicted mouse amino acid sequence reported here, with those of the relevant portions of the mouse gene Scn10a (36) and both rat and human Nav1.8 cDNAs (Refs. 4, 11, and 43; see “Discussion”).
We have isolated a novel Nav1.8 cDNA isoform in which there is skipping of a CAG trinucleotide at the boundary of Scn10a exon 16 and 17 transcribed sequences (36) (Fig. 3A). In accordance with the suggested nomenclature for alternatively spliced voltage-gated sodium channels, the CAG-skipped isoform will hereafter be referred to as Nav1.8c (Ref. 2, see “Discussion”). The frequencies of the Nav1.8c isoform among Nav1.8 cDNA clones from neonatal (P1), adult-control, and adult-axotomy DRG were very similar: respectively 14/20, 14/20, and 13/19. This suggests that Nav1.8c is not a minor isoform in DRG and that the Nav1.8c/Nav1.8 ratio is not under apparent regulation either during postnatal development or in the adult following axotomy.
A single example of another isoform, Nav1.8d, was also found from neonatal DRG, which had apparently mis-spliced at the exon17/intron 17 boundary leading to the retention of the 5′-nucleotides GTAG of intron 17 (Fig. 3B). This results in a frameshift with an open reading frame extending for only a further four amino acids and potentially encoding a two-domain channel (see Refs. 37 and 44), though mRNAs containing premature translation termination (nonsense) codons are apparent targets for rapid turnover by nonsense-mediated decay (45).
To test for tissue-specific regulation of differential splicing, Nav1.8 cDNA was also cloned from mouse trigeminal ganglia. All sequenced clones had the same 16 nucleotide substitutions found in DRG, and the frequency of the Nav1.8c isoform was 13/20, similar to that in DRG. The cloning method resulted in additional 5′-sequence (see “Experimental Procedures”) within which another nucleotide substitution was present in all 20 clones (T2905C), resulting in a predicted alanine residue as found in rat, dog, and human Nav1.8 (Table I and Fig. 2; Ref. 46).
Nav1.8 is transcribed from the Scn10a gene (2). In order to confirm the absence of both the previously described exon 17-transcribed sequence (36) and of differential splicing of such a 141-nt exon, three overlapping genomic PCR clones were used to span the 4.1 kb from the middle of Scn10a exon 16 to the 5′-end of exon 18 (sequence deposited under GenBank™/EBI accession number AJ622906). After PCR with primers derived from the published sequence flanking exon 17,20/20 genomic clones incubated with Bsu36I were uncut (data not shown), as opposed to being linearized due to the recognition site in the published cDNA sequence. DNA sequences of exon 17 genomic clones each confirmed the 12 base changes expected from the cDNA sequence we have reported above.
Scn10a intron 16 is 1561 nt and intron 17 is 2156 nt long, similar to the respective reported estimates of 1.4 kb and 2.3 kb (36). Neither intron contains an additional exon 17-related sequence, and the clones containing the exon 16 sequence each confirmed the four base changes expected from the cDNA sequence. Repetitive elements are present in both introns, as described in AJ622906, including a B4 element (47) within intron 17 with the adjacent dinucleotide repeat (GT)2GA(GT)21AT(GT)4AT(GT)3 that is a potential polymorphic intragenic marker (respectively, nt 2799–3063 and 3064–3129).
The published rat cDNA sequences (4, 11) both have a CAG trinucleotide corresponding to the start of mouse exon 17 (36), i.e. they are analogous to the mouse CAG-inclusive Nav1.8 isoform. RT-PCR of rat Nav1.8 from adult DRG and trigeminal ganglia resulted in the expected product sizes of 450 bp (data not shown). DNA sequence analysis of the cloned products revealed that only 3/20 DRG clones and 2/20 trigeminal ganglia cDNA clones were Nav1.8c sequences, with skipping of a CAG trinucleotide (Fig. 3A). Therefore, there is conservation of the Nav1.8c isoform in rat, with an apparent species-specific difference in the ratio of Nav1.8c to Nav1.8 transcripts in both DRG and trigeminal ganglia (each with a continuity corrected χ2 test, p = 0.001).
In the case of mouse Nav1.5, if there was analogous alternative splicing to that of Nav1.8, it would result in the introduction of an additional CAG trinucleotide by comparison to the published sequence (see “Discussion”). The wider tissue distribution of Nav1.5 compared with Nav1.8 allowed more tissues to be tested, and we investigated neonatal DRG and heart, together with adult heart, thigh muscle, and brain. The expected size of the Nav1.5 product is 378 bp, whereas that of 219 bp would correspond to the previously described mouse H1–2 and rat Nav1.5a isoforms (hereafter Nav1.5a), which have in-frame deletions of 159 bp, equivalent to exon 18 of human Nav1.5 (6, 37, 38).
The two expected sizes of RT-PCR products were detected in each of the tissues, with the apparent relative abundance of the larger isoform being: adult heart > neonatal heart > brain, neonatal DRG > thigh muscle (Fig. 4). The Nav1.5 product was extremely rare in normal adult thigh muscle, only being detected using the larger volume of input RT reaction tested. We show that the ratio of the two products varied very widely between tissues, with an apparent developmental change in heart between the neonate with similar levels of Nav1.5 and Nav1.5a, to adult with predominantly Nav1.5 isoform, whereas the smaller isoform was predominant in brain, thigh muscle, and neonatal DRG. DNA sequencing of smaller isoform cDNAs from each of the tissues confirmed them to be Nav1.5a.
DNA sequencing of ten large isoform mouse Nav1.5 cDNA clones from each of the five tissues did reveal an additional CAG trinucleotide at the predicted exon junction site in some cDNA clones from neonatal heart (2/10), adult heart (1/10), and adult thigh muscle (1/10) (Fig. 5). The identification of this isoform, designated Nav1.5c (see “Discussion”), establishes that differential splicing at the Scn5a intron 17/exon 18 junction produces abundant (CAG-skipped) Nav1.5 and rare (CAG-inclusive) Nav1.5c species.
Nav1.9 cDNA was cloned from neonatal mouse DRG, and 58 individual clones were tested for the presence of the single expected StuI or BbsI sites. The StuI recognition site overlaps the junction of exons 14 and 15A, and would not occur in a differentially spliced cDNA. No examples of alternative splicing were detected, as was confirmed by DNA sequencing of six Nav1.9 cDNA inserts (data not shown), which were identical to the previously published cDNA sequence (Ref. 39, AF118044).
As adult rat trigeminal ganglia is known to express both Nav1.8 and Nav1.9 mRNAs (11, 13), we investigated whether it also expresses Nav1.5 mRNA and whether a rat Nav1.5c isoform exists. RT-PCR of trigeminal ganglia did result in a product of the expected size, and DNA sequencing of individual Nav1.5 cDNA clones showed that 2/30 were of the Nav1.5c isoform that included an additional CAG trinucleotide (Fig. 5).
In order to optimize a quantitative RT-PCR assay for mouse Nav1.8 mRNA expression,2 we cloned a partial length Nav1.8 cDNA encoding part of the interdomain cytoplasmic loop IDII/III. This loop has the lowest sequence similarity to Nav1.5 or Nav1.9 and, along with the intracellular IDI/II loop, is the least conserved region of voltage-gated sodium channel α-subunits (11, 13, 46). By comparison to the published cDNA sequence derived from genomic DNA (36), Nav1.8 cDNA cloned by RT-PCR from mouse DRG and trigeminal ganglia have 12 base substitutions in the complete 141 nt of exon 17-transcribed sequence, plus 5 substitutions in exon 16-transcribed sequence (Table I). Among the 12 predicted amino acid differences, the sequence 1067QREESPRV includes a putative protein kinase C phosphorylation site motif (underlined), but this motif has a high probability of occurrence and is not conserved in rat or human predicted sequences (Fig. 2). The absence of a mutually exclusive alternative exon 17 with identity to the published sequence (36) was demonstrated by cloning and sequencing 4.1 kb of the Scn10a gene from exons 16–18 (AJ622906). The base substitutions might be caused by the different mouse substrains used, 129/OlaHsd in this study compared with 129/SvJ (36), though we note that the rat Nav1.8 cDNA sequences from the outbred Sprague-Dawley (4, 11) and Wistar strains (this study) were identical. Recently, a mouse genomic sequence has been extended (working draft AC134403), which now encompasses 4.1 kb, and differs from it only by the insertion of (GT)3 within the dinucleotide repeat of intron 17.
A novel isoform of mouse Nav1.8 cDNA was isolated in which a CAG trinucleotide is skipped at the junction of exon 16 and 17 transcribed sequences (Fig. 3A). The 3′-codon of Scn10a exon 16 and the 5′-codon of exon 17 are each CAG, respectively encoding glutamine 1030 (Gln1030) and Gln1031 (36), but only the skipping of nt 3126–3128 from the 5′-end of exon 17 would result in canonical GT and AG dinucleotides at either end of the intervening intron, the so-called GT-AG rule (48). Therefore, it is most likely that this isoform results from skipping of the codon that encodes Gln1031 because of differential 3′-splice site selection (Fig. 6A). This isoform has been designated Nav1.8c, in accordance with suggested nomenclature (2), as it is the third alternatively spliced Nav1.8 cDNA sequence to be described. Previously, a cDNA sequence encoding an additional glutamine residue within the cytoplasmic loop IDI/II was isolated (4), which following the sequencing of the mouse gene (36), can be viewed as resulting from alternative splicing at the intron 11/exon 12 junction, and an unusual trans-spliced product has also been isolated (SNS-A, Ref. 49).
The sampling of mouse Nav1.8 cDNA sequences (n = 59) does not currently support major differences in the ratio of Nav1.8c (CAG-skipped) to Nav1.8 (CAG-inclusive) mRNA between either neonatal and adult DRG, or between adult intact and postaxotomy DRG. The Nav1.8c isoform accounts for approximately two-thirds of Nav1.8 cDNA clones from mouse DRG. As trigeminal ganglia also express Nav1.8 (11), we checked for the possibility of tissue-specific regulation of splicing, especially as neurons of the trigeminal ganglia have a mixed origin from both the neural crest and ectodermal placodes, in contrast to trunk DRG, which originate entirely from migratory cells of the neural crest (50). Nav1.8c was also present in trigeminal ganglia, and its ratio to Nav1.8 was similar to that in DRG. The Nav1.8c isoform was conserved in the rat (Figs. (Figs.3A3A and and6A),6A), but occurred at the statistically lower frequencies of 3/20 DRG and 2/20 trigeminal ganglia Nav1.8 cDNA clones (p = 0.001).
The alternative splicing of Scn10a is unusual, because the mouse Nav1.8 isoform is spliced between a conservative intronic CAG and an immediately downstream exonic CAG trinucleotide, a so-called CAG/CAG motif (51), whereas the Nav1.8c isoform results from splicing after the downstream CAG trinucleotide. Generally, the first AG dinucleotide downstream of the intron branchpoint sequence (BPS) is most likely to be used as the 3′-splice site (52), there being a dramatic avoidance of an AG dinucleotide upstream of the 3′-splice site AG, with the average nearest 5′-AG around 30 nt distant (48, 52, 53). Alternative splicing of Nav1.8 in rat also results from differential 3′-splice site selection at a CAG/CAG motif, determined from a genomic clone (working draft AC135877) identified as including SCN10A exon 15–20 sequences, and results in the skipping of the codon for a glutamine residue numbered either 1031 (4) or 1032 because of the presence of an additional proline residue 521 in the latter sequence (11). Alternative splicing of a trinucleotide at a CAG/CAG motif also occurs in the mouse serine protease adipsin (formerly 28K) and sterolin-2, human E2A transcription factors E12/E47, Xenopus transcription elongation factor TFIIS.oB, and chicken very low density apolipoprotein II (51, 54-57).
In marked contrast to mouse, there is a bias against the Nav1.8c isoform in rat, although intron 16 sequences are highly conserved between the two species (data not shown) with the 3′-splice sites identical from nucleotide positions −27 to +4 (positive numbers indicating exon nt; AJ622906, nt 1743–1773). However, immediately upstream there is a less conserved 61 nt aligning to 33 nt in mouse (AJ622906, nt 1710–1742). Differential 3′-splice site selection will also be influenced by trans-acting factors, e.g. in a human GαS GTP-binding protein minigene, the variant CTG/CAG motif is biased to the CAG-inclusive isoform by SF2/ASF (splicing factor 2/alternative splicing factor) and to the CAG-skipped isoform by the antagonistic splicing factor hnRNP A1 (heterogeneous nuclear ribonucleoprotein A1) (58). Marked species-specific differences in the ratio of alternative splicing products have previously been documented between mouse and rat (59, 60).
In the case of human Nav1.8, no skipping of a CAG trinucleotide has been reported, though the cDNA sequence includes the relevant CAG triplet encoding Gln1030 (43), and the intron 16/exon 17 junction sequence CACTAGCAGGAG (AC137625, nt 60282–60271; exon 17 underscored) infers that a putative human Nav1.8c isoform would utilize a TAG/CAG motif. Such a variant motif is utilized in alternative splicing of rat prolactin (61), and closely spaced 3′-splice site AG dinucleotides compete in in vitro splicing with a hierarchy of CAG = TAG > AAG > GAG (62).
The Scn11a product Nav1.9 was not expected to undergo analogous alternative splicing, as there is no CAG/CAG-like motif in the intron 14/exon 15A junction sequence tgtcagGCCTGT (39), and no alternative splicing of Nav1.9 was detected in neonatal DRG by either restriction enzyme analysis or DNA sequencing. It is noteworthy that Nav1.8 and Nav1.5 proteins are more closely related to one another than to Nav1.9 (1, 2), with the IDII/III loop of Nav1.9 as the domain with by far the lowest similarity (13).
Nav1.5 mRNA was readily detected in heart, as expected, with a lower apparent expression in brain (5-8, 41, 63) and neonatal mouse DRG (Fig. 4A). Expression in embryonic rat DRG has recently been detected utilizing a combined RT-PCR and restriction enzyme polymorphism assay, with levels decreasing during development from embryonic day 15 to little or no detectable Nav1.5 at postnatal day 0 (9). Nav1.5 mRNA expression in normal adult thigh muscle was extremely rare (Fig. 4B). Using less sensitive methods, it has previously been either not detected (8, 63, 64) or found only in trace amounts in adult skeletal muscle (65), though it is abundant following surgical denervation (8, 41, 65). The expression in normal skeletal muscle may be because of satellite cells (66). When compared with the previously published mouse Nav1.5 cDNA sequence (Ref. 37, AJ271477), all sequenced clones contained the silent nucleotide substitution G3284A, plus the two substitutions G3448C and C3449T, which together result in the conservative predicted amino acid change S1133T. A mouse genomic clone also has these three substitutions (AC121922; nt 56139, 60390, and 60391), and a predicted threonine residue occurs at the corresponding position in both rat (6, 8, 41) and human Nav1.5 (38, 63).
Unexpectedly, we have found that the ratio of Nav1.5 to the smaller Nav1.5a isoform is under apparent developmental regulation in mouse heart, with similar levels in the neonate as compared with predominantly Nav1.5 isoform in the adult (Fig. 4A). Previous reports have shown the Nav1.5 isoform to be more abundant in adult rat and mouse heart (6, 37), but we are not aware of neonatal or embryonic studies that have distinguished between Nav1.5 mRNA isoforms. Differences in gene expression between neonatal and adult heart (67) include the onset of expression at postnatal day 15 of the auxiliary β2-subunit, which covalently interacts with Nav1.5 (68).
We anticipated that a novel alternative splicing of mouse Nav1.5 mRNA might occur, with the introduction of a CAG trinucleotide, as the mouse genomic DNA sequence included TTT CAG CAG GAA TCC CAA (AC121922, nt 57420–57437) i.e. an intronic putative CAG/CAG motif upstream of cDNA sequence corresponding to exon 18 (underlined; Refs. 37 and 38). Whereas such a putative CAG codon (bold type, above) is absent in rat and mouse published cDNAs (8, 37, 41), the human Nav1.5 cDNA does include a CAG codon transcribed from the start of exon 18 (hH1, Refs. 38 and 63; Fig. 6B). A rare Nav1.5 isoform with an additional CAG trinucleotide was indeed isolated from mouse neonatal heart, adult heart, and thigh muscle (Fig. 5). As with mouse Nav1.8, this alternative splicing is caused by differential 3′-splice site selection at a CAG/CAG motif, with the CAG-skipped isoform favored. We have designated the rare CAG-inclusive isoform Nav1.5c, because it is the third alternatively spliced isoform of rodent Nav1.5. A rat isoform with a deletion of 159 bp equivalent to exon 18 was designated Nav1.5a (6, 38), whereas the analogous mouse isoform was referred to as H1–2, with an apparently non-functional H1–3 isoform having an in-frame deletion corresponding to exons 17 and 18 of human SCN5A (37, 38). The primers used in the present work do not detect the H1–3 isoform.
The conservation of a rare Nav1.5c isoform in rat was confirmed in adult trigeminal ganglia (Figs. (Figs.55 and and6B),6B), and we are not aware of a previous demonstration of Nav1.5 expression in this tissue. This finding predicts the presence of a tetrodotoxin-resistant sodium current with fast activation and inactivation kinetics in trigeminal ganglia, the sensory neurons of which innervate the oromaxillofacial region (9, 50). Within the IDII/III cytoplasmic loop of mouse and rat Nav1.5 there is a putative site for phosphorylation by CK2 (previously known as casein kinase 2), the minimum consensus sequence (S/T)XX(E/D) of which (69) corresponds in mouse to 1075ESSKQ/ESQ, where the forward slash indicates the junction of residues encoded by exons 17 and 18 (Fig. 6B). However, this putative phosphoacceptor site is destroyed by the introduction of a glutamine residue in the Nav1.5c isoform, which raises the possibility of alternative splicing regulating Nav1.5 phosphorylation. Such a mechanism could apply equally to the alternatively spliced rat (this report) and human isoforms. A data base search for a human CAG-skipped Nav1.5 cDNA identified the two recent sequences HJSCN5A, expressed in jejunal circular smooth muscle (70), and heart hH1b (71), the latter work also mentioning a non-data base clone hH1a (72). Most recently, the CAG-skipped isoform has been reported to make up 65% of Nav1.5 transcript in human hearts (73).
In summary, we have demonstrated the conservation of an unusual alternative splicing mechanism by the mouse Nav1.8 and Nav1.5 sodium channels. Alternative 3′-splice site choice at CAG/CAG motifs result in Nav1.8c/Nav1.8 and Nav1.5/Nav1.5c isoform pairs, which are also conserved in rat and differ, respectively, by the absence or presence of a glutamine residue within the IDII/III cytoplasmic loop. Single amino acid variants within a sodium channel IDII/III loop can alter electrophysiological properties and have been implicated in disease. The conservation of these novel isoforms suggests they could have important functions, especially as Nav1.5 is the principal cardiac sodium channel, and Nav1.8 has been shown to play a major role in nociception.
We thank Dr. Linda Hunt (Institute of Child Health, University of Bristol) for advice on statistics and Robert Pope for help with the figures.
The nucleotide sequence(s) reported in this paper has been submitted to the GenBankTM/EBI Data Bank with accession number(s) AJ623269 (mouse Nav1.8 cDNA), AJ623270 (mouse Nav1.8c cDNA), AJ622906 (mouse Scn10a exons 16–18), AJ623271 (rat Nav1.8c cDNA), AJ623272 (rat Nav1.5c cDNA), AJ623273 (mouse Nav1.5c cDNA), and AJ623274 (mouse Nav1.5 cDNA).
*This work was supported by the Medical Research Council (MRC) and the Biotechnology and Biological Sciences Research Council (BBSRC).
1The abbreviations used are: DRG, dorsal root ganglia; RT, reverse transcription; nt, nucleotides; aa, amino acid.
2N. Kerr, F. Holmes, and D. Wynick, manuscript in preparation.