Neuron,

Vol. 4, 233-242,

February,

1990, Copyright

0 1990 by Cell Press

Primary Structure and Expression of a Sodium Channel Characteristic of Denervated and Immature Rat Skeletal Muscle Roland Liquiong and

G. Kallen:+ Zu-Hang Sheng:+ Chen,*+ Richard B. Rogart,§

Robert

Jane

Yang:*

1. Barchi**

*David Mahoney Institute of Neurological Sciences +Department of Biochemistry and Biophysics *Department of Neurology University of Pennsylvania School of Medicine Philadelphia, Pennsylvania 19104 SDepartment of Medicine University of Chicago Chicago, Illinois 60637

Summary The a subunit of a voltage-sensitive sodium channel characteristic of denervated rat skeletal muscle was cloned and characterized. The cDNA encodes a 2018 amino acid protein (SkM2) that is homologous to other recently cloned sodium channels, including a tetrodotoxin (TTX)sensitive sodium channel from rat skeletal muscle (SkMl). The SkM2 protein is no more homologous to SkMl than to the rat brain sodium channels and differs notably from SkMl in having a longer cytoplasmic loop joining domains 1 and 2. Steady-state mRNA levels for SkMl and SkM2 are regulated differently during development and following denervation: the SkM2 mRNA level is highest in early development, when TTX-insensitive channels predominate, but declines rapidly with age as SkMl mRNA increases; SkM2 mRNA is not detectable in normally innervated adult skeletal muscle but increases >lOO-fold after denervation; rat cardiac muscle has abundant SkM2 mRNA but no detectable SkMl message. These findings suggest that SkM2 is a TTX-insensitive sodium channel expressed in both skeletal and cardiac muscle. Introduction

The action potential in skeletal muscle is produced by voltage-sensitive sodium channels that are functionally very similar to those in nerve. These channels can be distinguished, however, by aspects of their gating kinetics, their sensitivity to neurotoxins, and their reactivity with specific immunoreagents (Campbell and Hille, 1976; Cruz et al., 1985; Haimovich et al., 1987). Sodium channels isolated from skeletal muscle resemble those from brain and eel in containing a large a subunit (molecular mass 230-270 kd) that is extensively modified at the posttranslational level (see Catterall, 1986; Barchi, 1988, for reviews). Muscle and brain sodium channels also contain one or two smaller p subunits (37-39 kd), but oocyte expression studies indicate that the a subunit contains all of the essential structural information necessary to form functional voltage-dependent ion channels (Noda et al., 1986b; Goldin et al., 1986; Stuhmer et al., 1987; Suzuki et al., 1988; Auld et al., 1988; Trimmer et al., 1989).

In addition to tissue-specific forms of the sodium channel, more than one sodium channel subtype is expressed in both nerve and muscle. Three different sodium channel a subunits have been cloned from adult rat brain (Noda et al., 1986a), although none a@ pear to be expressed at significant levels in muscle (Cooperman et al., 1987). In muscle, multiple subtypes have been distinguished by their reactivity patterns to antibodies (Haimovich et al., 1987) and to a variety of channel neurotoxins (Redfern and Thesleff, 1971; Harris and Thesleff, 1971; Rogart and Regan, 1985; Gonoi et al., 1987) and by their electrophysiological properties (Paponne, 1980; Weiss and Horn, 1986). Notably, sodium channels in adult innervated skeletal muscle are blocked by nanomolar concentrations of tetrodotoxin (TTX) and saxitoxin (STX), whereas those in denervated skeletal muscle as well as those in normal cardiac muscle (Brown et al., 1981) are resistant to these toxins. The primary structures of the a peptides of sodium channels from eel and rat brain (RBSC-I, -II, and -Ill) and of one channel subtype from rat skeletal muscle (SkMl) have been determined from the nucleotide sequences of full-length cDNA clones (Noda et al., 1984, 1986a; Kayano et al., 1988; Trimmer et al., 1989; Auld et al., 1988). A partial sequence for a Drosophila sodium channel has also been obtained (Salkoff et al., 1987). In each case the a subunit contains four large (~275 amino acids) regions of internal homology, which are highly conserved between species. Within each of these internal repeat domains are six to eight putative transmembrane helices, one of which (S4) is highly amphipathic, with a positively charged residue in every third position. Several models of channel tertiary structure have been proposed (Noda et al., 1984; Greenblatt et al., 1985; Guy and Seetharamulu, 1985) in which the four repeat domains of one a subunit are organized within the membrane in a pseudosymmetric fashion to form a single ion channel. The first muscle-specific sodium channel to be cloned (SkMl) is expressed in both innervated and denervated adult muscle and is sensitive to nanomolar concentrations of lTX and p-conotoxin (Trimmer et al., 1989). We now report the isolation and sequence of cDNA clones encoding the a subunit of a second novel sodium channel from rat skeletal muscle, SkM2. The primary sequence of SkM2 exhibits no greater homology to SkMl than to RBSC-I, -II, or -III. Expression studies indicate that SkM2 mRNA is not detectable in normally innervated skeletal muscle, but appears in abundance following denervation. Furthermore, unlike SkMl mRNA, probes specific for the SkM2 message hybridize with a similarly sized transcript that is prominently expressed in cardiac muscle and in the muscle-derived cell line L6. SkM2 mRNA is expressed early in the postnatal development of skeletal muscle and declines with age as the expres-

sion of SkMl increases. These observations are consistent with the concept that SkM2 is a TTX-insensitive form of the muscle sodium channel. Results Isolation and Analysis of SkM2 cDNAs Denervated skeletal muscle expresses both a TTXsensitive and a lTX-insensitive voltage-dependent sodium channel (Redfern and Thesleff, 1971; Rogart and Regan, 1985). Northern blot and nuclease protection studies have shown that the predominant sodium channel mRNAs in rat skeletal muscle are homologous, but not identical, to any of the three sodium channel messages previously characterized in rat brain (Cooperman et al., 1987). Using a cDNA probe encoding the relatively conserved fourth internal repeat domain (D-4) from RBSC-II (pRB211, 1151 bp, nucleotides 4794-5945, amino acids 1529-1912; Cooperman et al., 1987), cDNAs encoding two different a subunits were isolated from denervated skeletal muscle cDNA libraries. The complete characterization of one of these a subunits, designated SkMl, has recently been reported (Trimmer et al., 1989). This muscle sodium channel subtype is present in both innervated and denervated muscle and has been shown to be sensitive to nanomolar concentrations of TTX and p-conotoxin in an oocyte expression system. We report here the complete sequence of the second muscle a subunit subtype, designated SkM2, which is expressed in adult skeletal muscle only following denervation. Overlapping cDNA clones containing the remainder of the SkM2 coding sequence were obtained through a series of library screenings using progressively more 5’and 3’directed probes. These clones have been designated pSkM2A, pSkM2-B, and pSkM2-C. Their size and location are shown in Figure 1. The inserts were mapped with restriction enzymes, reciprocal Southern blot analysis, and nucleotide sequencing of the 5 and 3’ termini by the dideoxy method on singleor double-stranded templates (Sanger et al., 1977; Maniatis et al., 1982; Ausubel et al., 1987). The establishment of the identity, orientation, and approximate location of these cDNAs was assisted initially by comparison with the nucleotide sequences of other sodium channels. The nucleotide sequence of each clone was then determined in its entirety on both the coding and the noncoding strands using sequential polar deletions prepared by the exonuclease Ill/mung bean exonuclease method. The complete nucleotide sequence consists of 7076 bases (Figure 2). It includes an open reading frame of 2018 amino acids that encodes a protein with a calculated molecular mass of 227,339 daltons. A 206 bp 5’ untranslated region precedes the first in-frame ATG initiation codon. Two additional ATC codons exist further 5’; one of these codons appears to be invariant (-8 to -6) in known sodium channel sequences (Trimmer et al., 1989). Despite the fact that the sequence directly 5’to the trans-

3’UT 816

j

bp

pSkMZ-B i

pSkMZ-C

Figure 1. Overlapping cDNA Clones Constituting SkM2 Designation of individual clones used in the text are as follows (nucleotide 1 is defined as the first nucleotide in the most 5 clone): pSkM2A = 3031 bp, I-3031, 5’ untranslated region, and amino acids l-942; pSkM2-B = 3354 bp, 2102-5456, amino acids 633-1751; pSkM2-C = 1752 bp, 5324-7076, amino acids 1707-2018, and 3’ untranslated region. The initiation codon is at position 206-209, and the termination codon is at 6260-6262 in the complete SkM2 sequence.

lation initiation codon, -ATGAGAAG.ATG.G-, bears only modest resemblance to the consensus sequence established for eukaryotic initiation sites, it does contain the most important features of the consensus sequence, namely, an A residue at -3 and a G residue at +4 (Kozak, 1984). A termination codon, TAG, at 6254-6256 (coordinates as given in Figure 1) results in a 3’ untranslated region of 816 bp that apparently does not extend to include either a consensus polyadenylation signal site (AATAAA) or a poly(A) tail (Proudfoot and Brownlee, 1976). The internal organization of the amino acid sequence for SkM2 is similar to that for SkMl (Figure 3). As with all other sodium channels, SkM2 contains four internal domains (D-l to D-4) that share42%-52% sequence homology. Each domain in SkM2 exhibits a characteristic hydrophobicity profile comparable to that interpreted to include six to eight membrane-spanning helices in other sodium channels (Creenblatt et al., 1985; Noda et al., 1986a; Guy and Seetharamulu, 1986). Each domain also contains an amphipathic helix (S4) with highly conserved positively charged residues in every third position; these putative helices are homologous in location and sequence to those in SkMl and other previously described sodium channels. The four homologous domains in SkM2 are related to, but distinct from, those of other sodium channels reported previously; for corresponding domains in other mammalian sodium channels, homologies with SkM2 range between 72% and 80% (Table 1): the homology to eel and Drosophila channels in these regions is slightly lower. Surprisingly, the homology between corresponding domains in SkM2 and SkMl is no greater than that between SkM2 and the three rat brain sodium channels. The interdomain region linking D-3 and D-4 (lD3-4) is highly conserved between SkM2 and SkMl with only 4 substitutions in 48 amino acids, but, again, even stronger homology is found in this region between SkM2 and the rat brain channels. SkM2 differs from SkMl in several significant aspects. First, the putative cytoplasmic loop connecting

Rat Skeletal 235

Muscle

Sodium

Channel

Expression

2002 1820

diadfPP----spdrdr::iy /I ppsssPPqgqtvrpwk~~Ib

Figure

3. Comparison

II ,

of Amino

Acid

Sequences

of SkM2

and

SkMl

SkM2, upper, II; SkMl, lower, I. Identical residues are denoted by vertical bars. The regions encompassing the four internal repeat domains are shown by vertical lines in the right margin. The suggested locations of six transmembrane helices within each domain, analogous to those in the model of Noda et al. (19861, are designated by horizontal lines. Potential secondary modification sites were located by consensus sequence analysis and are indicated as follows: open squares, glycosylation; circled stars, CAMP phosphorylation. Symbols for sites present in both channels are filled.

Rat Skeletal 237

Table

Muscle

1. Homologies

Sodium

Channel

between

Expression

Amino

Acid

Sequence

of SkM2

and

Other

Sodium

Channels

SkM2

SkMl RBSC-I RBSC-II RBSC-III Eel D. melanogaster

N

D-l

ID1-2

D-2

ID2-3

D-3

ID3-4

D-4

C

58 53 53 55 45

72 75 74 76 61 42

14 44 40 33 12

80 78 80 80 74 52

23 27 27 26 16

80 77 79 79 72 52

83 91 89 89 76

74 77 77 75 71 51

51 53 54 53 34

Homology (expressed as a percent) was calculated considering only exact matches and excluding conservative substitutions. tions were considered as a single nonidentity independent of length. SkMl (Trimmer et al., 1989); RBSC-I (Noda et al., 1986a); RBSC-II (Noda et al., 1986a); RBSC-III (Kayano et al., 1988); eel (Noda 1984); D. melanogaster (Salkoff et al., 1987).

1

2

3

4

5

6

i’

-9.5 - 7.5

-4.4

- 2.4

.:: -1.4

1

B

2

‘,’ -9.5 - 1.5 -4.4

-2.4

- 1.4

Figure 4. Northern Steady-State mRNA

Blot Analysis Showing the Distribution of Levels for SkM2 and SkMl in Various Tissues

Total RNA (20 ug) was subjected to denaturing 1% agarose gel electrophoresis, electrotransfer, hybridization with antisense [32P]RNA probes (-IO9 cpm/ug), and autoradiography. Subtypespecific probes (numbered from the first nucleotide of the Suntranslated region for each sequence) were as follows: SkMl = nucleotides 3045-3276; SkML = nucleotides 6209-7076. Comparable results were obtained with an SkM2-specific probe to a portion of the ID2-3 coding region encompassing nucleotides 3220-3586. (A) SkMZ-specific probe. (B) SkMl-specific probe. Lane 1, innervated adult skeletal muscle; lane 2, denervated skeletal muscle; lane 3, brain; lane 4, heart; lane 5, liver; lane 6, kidney; lane 7, uterus. Innervated muscle RNA was obtained from adult lower leg (anterior tibia1 and gastrocnemeus-soleus muscle), and denervated adult muscle was from the same muscle groups 5 days after the muscles were denervated by transection and removal of a portion of the sciatic nerve in the upper thigh. Numbers at the right of each panel indicate the location of size standards (in kilobases) run on the same gels (Bethesda Research Laboratories).

Deleet al.,

D-l and D-2 is 172 amino acids longer in SkM2 and in this regard is similar in size to the ID1-2 of the rat brain channels. Second, the predicted extracellular loop connecting putative transmembrane helices S5 and S6 in D-l is considerably shorter in SkM2 than in SkMl. A number of the consensus N-glycosylation sites present in this loop in SkMl are absent in the SkM2 sequence (see Figure 3).

Tissue Distribution mRNA Transcripts

of SkMl and SkM2

Total cellular RNA isolated from various tissues was size-fractionated on denaturing agarose gels, transferred to nylon membranes, and probed with SkMland SkM2-specific antisense a*P-radiolabeled transcripts. In adult rat, probes specific for SkMl hybridized to ~8.5 kb transcripts in total RNA preparations from skeletal muscle but not to RNAs from other tissues examined, including brain, heart, liver, kidney, and uterus (Figure4). An 8.5 kb transcript hybridizing to the SkM2 probe was prominent in the heart but was not detected in other adult tissues. In the adult rat the steady-state level of SkMl mRNA has been shown to increase slightly in skeletal muscle following surgical denervation (Cooperman et al., 1987). SkM2 mRNA is not detectable in Northern blots of adult innervated muscle, but it is expressed abundantly in the same muscle within 5 days following denervation (Figure 4). Thus while both SkMl and SkM2 sodium channel mRNAs are present in denervated muscle, the only mRNA detectable with these probes in innervated muscle is SkMl,

Expression of Sodium Channels Muscle and 16 Cells in Culture

in the Neonatal

RNA transcripts for both SkMl and SkM2 are detectable at low levels in the RNA preparations obtained from neonatal skeletal muscle. SkMl sodium channel message increases about IO-fold between postnatal day 1 and day 35, with the most rapid increase being observed after day 5 (Figure 5). In contrast, the steadystate level of SkM2 message decreases during muscle development, being highest at 1 day following birth and declining to undetectable levels at 35 days postpartum. In the same samples, a actin mRNA levels

Neuron 238

A

157

SkM

2

SkM

1

14

35

- 1.4 B % maximum

Figure 6. Amounts of SkM2 and SkMI mRNAs in 16 Muscle at 4 and 13 Days in Tissue Culture following Passage under ditions That Inhibit Fusion

Cells Con-

Lanes 1 and 2, SkM2-specific probe at 4 and 13 days; lanes 3 and 4, SkMl-specific probe at 4 and 13 days in culture. Numbers to the right of the panel indicate the location of size markers on the same gel.

SkM2

transcript;

remains time

the

constant

the

SkM2

ratio

from

of

SkM2

to

to

days,

birth

mRNA

levels

can

14

fl actin after

no longer

mRNA which

be reliably

quantitated.

The relative were examined which inantly 0 10

0

20

30

40

Days Post Partum -+SkM Figure

5. Expression

1

of Sodium

fSkM

2

Channels

during

Development

(A) Developmental time course of levels of steady-state mRNA for SkMZ, SkMl, and actins. Each row is a section of a Northern blot of total RNA electrophoretically separated and transferred as in Figure 4. Top row: blot was hybridized with SkM2 antisense 13*P]RNA; lanes contain 20 pg of total RNA from (left to right) 1,5, 7,14, and 35 day postpartum rat skeletal muscle. Middle row: the same samples probed with SkMl antisense [32P]RNA. Bottom row: the same RNA samples probed with antisense actin r2P]RNA demonstrating both a and p actin transcripts. Numbers to the right of each row indicate the size (in kilobases) of the labeled band as determined from standards run on the same gel. (B) Quantitation of the SkMl and SkM2 messages in the RNA samples shown in (A). Bands were scanned densitometrically, and the results are expressed as percentage of maximum observed present at each time point. Circles, SkMl; triangles, SkM2.

remain

approximately

postpartum, tonically. with

while The

a time

rate course

constant [3 actin

between

message

of decline comparable

of

levels 0 actin to

that

days decline

mRNA seen

1 and

35

mono-

occurs for

the

has

amounts in the

previously

TTX-insensitive

of SkMl and muscle-derived

been

shown

sodium

SkM2 cell

mRNAs line L6,

to express channels

predomin

culture

(Stallcup and Cohn, 1976; Lawrence and Catterall, 1981; Haimovich et al., 1986). Northern blots of total RNA derived from L6 myoblasts were probed with SkMl and SkM2-specific probes: no mRNA was detectable with the SkMl probe, whereas the SkM2 probe hybridized strongly to an 8.5 kb message (Figure 6). Discussion Voltage-dependent

sodium

channels

appear

to

con-

stitute an extended multigene family in the rat. Three different isoforms of the sodium channel have been sequenced and characterized in rat brain, but none of these are present at detectable levels in rat skeletal muscle. Recently, an additional sodium channel (SkMZ) has been cloned from skeletal muscle and sequenced (Trimmer

et al., 1989).

When

the

SkMl

full-length

mes-

sage was expressed in Xenopus oocytes, the resultant voltage-dependent sodium channel was sensitive to nanomolar concentrations of TTX and p-conotoxin, consistent with it being the predominant sodium channel found in adult innervated muscle (Trimmer et al., 1989). However, the existence of other sodium channel subtypes in skeletal muscle was anticipated given the known reactivity patterns of muscle sodium

Rat Skeletal 239

Muscle

Sodium

Channel

Expression

channels to neurotoxins and monoclonal antibodies. We have now isolated an additional member of the sodium channel multigene family, SkM2, which appears to be characteristic of denervated muscle. The most striking aspect of the SkM2 channel is its expression during development and following denervation in adult skeletal muscle. SkM2 mRNA is undetectable in innervated adult muscle but increases at least IOO-fold in the same muscle groups following surgical denervation. In this respect the SkM2 message behaves in a manner consistent with a message encoding the physiologically defined TTX-insensitive form of the muscle sodium channel that appears only following denervation in adult muscle. In developing rat skeletal muscle, the TTX-insensitive sodium channel isoform is present in the perinatal period, whereas the T-TX-sensitive adult isoform appears gradually over the first 3 weeks of life (Harris and Marshall, 1973; Sherman and Catterall, 1982; Lombet et al., 1983). When SkMl and SkM2 steady-state mRNA levels are determined over this time period, we find that SkM2 mRNA is at its highest level in neonatal muscle and decreases to an undetectable level after 15 days of age, whereas SkMl mRNA is barely detectable at birth and rises dramatically with age in the same time period. In this respect also, the SkM2 message behaves as expected for the TTX-insensitive isoform of the skeletal muscle sodium channel. The muscle-derived cell line L6 expresses predominantly TTX-insensitive sodium channels in culture (Stallcup and Cohn, 1976; Lawrence and Catterall, 1981; Haimovich et al., 1986). As expected if SkM2 were to encode a TTX-insensitive isoform, the SkM2 message is easily identified on Northern blots of L6 cell RNA, whereas no message is detected with the SkMl-specific probe. Finally, subtype-specific probes for the SkM2 message hybridize strongly with RNA from cardiac muscle, a tissue that is also characterized by the presence of a TTX-insensitive sodium channel. Thus, although confirmation must await the functional expression of a full-length clone, SkM2 represents a likely candidate for the T-TX-insensitive form of the muscle sodium channel.

domains (231-327 residues in length), each with ,hydrophobicity profiles indicating six to eight putative transmembrane a helices (-22 amino acids in length); an amphipathic positively charged helix in each domain (S4); two interdomain regions of low homology (291 and 263 residues in length); and one interdomain (ID3-4) with a very highly conserved sequence. The differences in the extent of homology in the various regions of SkM2 relative to other sodium channels (Table 1) indicate that, surprisingly, SkM2 is no more homologous to SkMl than to brain channels. Indeed, it appears that SkM2 more resembles RBSC-I and RBSCII in length and RBSC-III in sequence than it does SkMl. Overall, the hydropathicity profile of SkM2 is not significantly different from those of other channels, and therefore the previous models of sodium channel tertiary structure are equally relevant to SkM2 (Greenblatt et al., 1985; Noda et al., 1986a; Guy and Seetharamulu, 1986). At 2018 amino acid residues, the SkM2 protein is the largest sodium channel identified to date. By far the greatest contribution to the size differences among sodium channels results from variability in the length of the interdomain region separating domains 1 and 2 (IDI-2). This region differs in size among sequenced channels by up to 214 amino acid residues, and SkM2 possesses 172 more amino acids in this region than SkMl (Table 2). Comparison of the sequences of the various sodium channels in the ID1-2 region indicates that they can be classified into two groups. One includes the eel and rat muscle SkMl channels, which have short interdomains (121 and 154 amino acid residues); the second includes SkM2 and the three rat brain clones, all of which have long interdomains (>275 amino acids). Differences among channels in the length of the ID2-3 segment of up to 57 amino acid residues are also seen. SkM2 is 38 amino acid residues longer than SkMl in this region. Although highly conserved between SkMl and SkM2 (84% homology), the 54 amino acid ID3-4 segment of SkM2 is most homologous to that of RBSC-I and is more closely related to all three of the brain channel sequences than to SkMl. This region in SkM2 contains all II of the conserved lysine residues seen in other channels (Trimmer et al., 1989). Another region of divergence between SkMl and SkM2 that may have physiological significance occurs in the putative extracellular loop joining helices 5 and

Structure of SkM2 The SkM2 amino acid sequence exhibits the general features of previously characterized sodium channels, including four internally homologous repeat

Table

2. Segment

SkMl SkM2 RBSC-I RBSC-II RBSC-III Eel SkMl

(Trimmer

Acid

Sequences

N

Sizes

D-l

of Amino

ID1-2

D-2

ID2-3

D-3

ID3-4

D-4

C

Total

D-l -S5-6

126 127 123 124 123 117

327 289 302 303 303 307

121 293 335 324 277 154

231 231 231 231 231 231

223 261 221 220 217 204

269 270 270 270 267 270

54 54 54 54 54 54

249 248 ‘249 249 249 250

248 244 222 228 228 251

1840 2018 2009 2005 1951 1819

141 113 127 127 127 109

et al.,1989a);

RBSC-I

(Noda

of Various

et al., 1986a);

RBSC-II

Sodium

(Noda

Channels

et al., 1986a); RBSC-III

(Kayano

et al., 1988); eel (Noda

et al., 1984).

NeUrOll 240

6 in domain 1. SkM2 is 28 amino acids shorter than SkMl in this region. This loop is thought to be the major site for glycosylation in the SkMl channel a subunit (Trimmer et al., 1989), and the alteration in SkM2 results in a net loss of 3 of the potential glycosylation sites found in this loop in SkMl. In addition, it is likely from models of tertiary structure that this extracellular loop is located in the vicinity of the external opening of the ion pore, a region in which the neurotoxins TTX and STX are thought to bind. If SkM2 encodes the TTX-insensitive form of the muscle sodium channel, the S5-S6 loop of Dl may be a candidate for the site of STX and lTX binding to SkMl.

Consensus

Sites for Posttranslational

Modifications

The sodium channel a peptides purified to date are prominently glycosylated as well as phosphorylated, acylated, and sulfated (Catterall, 1986; Barchi, 1988). The deduced amino acid sequence of SkM2 predicts 20 asparagine-linked (N-linked) consensus glycosylation sites (Hubbard and Ivatt, 1981), 66 consensus Ser or Thr phosphorylation sites (5, 26, and 35 for CAMPdependent kinase, C-kinase, and casein kinase-II, respectively), and 1 tyrosine phosphorylation site between S5 and S6 of D-3. The potential glycosylation sites are predominantly in regions of the channel believed to lie on the extracellular side of the membrane (14 of 20); they tend to be clustered, especially in the long loop connecting putative helices 5 and 6 in domains 1 and 3, where there are 5 and 4 potential glycosylation sites, respectively. This differs from SkMl, in which 8 potential extracellular N-glycosylation sites are concentrated between S5 and S6 of D-l, with only 2 in the comparable loop of D-3. Of the 5 sites in the SkM2 D-I-S5-6 loop, 3 are conserved in SkMl, whereas 2 of the 4 sites in the SkM2 D-3-S5-6 loop are conserved between the two muscle channels. The longer ID5-6 loop of D-l in SkMl contains 5 tandem repeats of a 6-residue motif (Trimmer et al., 1989), which are absent from the SkM2 sequence. Of the 5 potential CAMP-dependent phosphorylation sites, 2 are in the ID1-2 cytoplasmic loop, a site shown to be phosphorylated in RBSC-II (Rossie et al., 1987). Two others lie in the cytoplasmic ID2-3 loop, and the last is in the amino-terminal portion of the molecule. None of the 5 phosphorylation sites are conserved between SkM2 and SkMl. Other features of the analysis of the SkM2 sequence are Ii’ potential myristylation sites, an amidation site (588), and a leutine-zipper pattern (916-937) (Vogt et al., 1987; LandSchulz et al., 1988), the significance of which remains unclear.

The Relationship Voltage-Dependent

of SkM2 to Other Sodium Channels

The SkM2 sodium channel occupies an intermediate position between the structure of the SkMl sodium channel of adult innervated muscle and the sodium channels of brain. In many respects, it appears more

closely related to the brain channels than to the other muscle channel. Determination of the significance of the structural differences noted between the muscle channels in regard to their physiological function and toxin binding awaits the outcome of in vitro expression studies with SkM2. To date these studies have proven difficult because of the nature of the restriction sites available for the production of a full-length clone, but these technical problems should eventually be amenable to alternate approaches. In the meantime, the clear differences in tissue and developmental expression of these two skeletal muscle sodium channel subtypes, along with the availability of the complete sequences for both, should prove valuable for further studies on the regulation of their expression. While the cloning of the SkM2 channel was in progress we became aware of work in progress, on a similar sodium channel from cardiac muscle (Rogart et al., 1989). This channel appears to be almost identical in sequence to the SkM2 channel reported here and almost certainly represents the message to which our SkM2 probes hybridize in total RNA preparations from cardiac muscle. The two sequences differ at only a small number of nucleotide positions, 3 of which are within the coding region of the sequences. These 3 differences do not result in amino acid changes and are present in only some cDNA clones. Thus they may be cloning artifacts. More detailed comparative work will be required to determine the origin and significance of these differences. Experimental

Procedures

Materials Guanidinium isothiocyanate, guanidinium hydrochloride, RNA molecular weight standards, phenol, oligo(dT), and nick translation kits were purchased from Bethesda Research Laboratories; Nytran membranes, from Schleicher & Schuell. In vitro trancription kits, RNAase-free DNAase, and RNAase inhibitor were from Stratagene Cloning Systems. [cGF’]lJTP (SP6 grade), [a-P32]dCTP, and cDNA cloning kits were from Amersham.

Generation and Screening of Muscle cDNA Libraries The cDNA library from IEday postdenervation adult skeletal muscle has been described previously (Trimmer et al., 1989). cDNA libraries from innervated adult skeletal muscle were generated either in our labs or by Stratagene Cloning Systems from poly(A)” RNA using both oligo(dT) and random hexameric (Pharmacia) primers for first strand cDNA synthesis with AMV reverse transcriptase. Second strand synthesis with RNAase t-l, DNA poll was followed by methylation, ligation of EcoRl linkers, cleavage by EcoRI, fractionation of inserts to be greater than 1 kb on a Sepharose 4B column, ligation of the inserts into EcoRI phosphatase-treated @IO arms, packaging, and infection of E. coli accordingto protocols and with reagents supplied by Amersham or Stratagene Cloning Systems. The libraries contained in excess of IO6 independent recombinants with a parental phage background of