Cloning and Functional Expression of a Voltage-gated Calcium Channel α1 Subunit from Jellyfish*

Voltage-gated Ca2+channels in vertebrates comprise at least seven molecular subtypes, each of which produces a current with distinct kinetics and pharmacology. Although several invertebrate Ca2+ channel α1 subunits have also been cloned, their functional characteristics remain unclear, as heterologous expression of a full-length invertebrate channel has not previously been reported. We have cloned a cDNA encoding the α1 subunit of a voltage-gated Ca2+ channel from the scyphozoan jellyfishCyanea capillata, one of the earliest existing organisms to possess neural and muscle tissue. The deduced amino acid sequence of this subunit, named CyCaα1, is more similar to vertebrateL-type channels (α1S, α1C, and α1D) than to non-L-type channels (α1A, α1B, and α1E) or low voltage-activated channels (α1G). Expression of CyCaα1 in Xenopus oocytes produces a high voltage-activated Ca2+ current that, unlike vertebrateL-type currents, is only weakly sensitive to 1,4-dihydropyridine or phenylalkylamine Ca2+ channel blockers and is not potentiated by the agonist S(−)-BayK 8644. In addition, the channel is less permeable to Ba2+than to Ca2+ and is more permeable to Sr2+. CyCaα1 thus represents an ancestral L-type α1 subunit with significant functional differences from mammalian L-type channels.

marily responsible for determining the pharmacology and physiology of the resulting current, and heterologous expression of cloned vertebrate ␣ 1 subunits has allowed correlation between molecular subtypes and native currents. L-type currents, sensitive to 1,4-dihydropyridines (DHPs), 1 are gated by the ␣ 1S subtype found in skeletal muscle (6) and the ␣ 1C (3) and ␣ 1D (7) subtypes expressed in heart, brain, and other tissues. DHP-insensitive, non-L-type ␣ 1 subunits include the ␣ 1B subtype, which is responsible for the -conotoxin GVIA-sensitive N-type current (8), the ␣ 1A subtype, which gates the -agatoxin IVA-sensitive P/Q-type current (9), and ␣ 1E , which gates R-type currents (10). In addition to these high voltage-activated (HVA) ␣ 1 subunits, the ␣ 1G subunit, responsible for the low voltageactivated (LVA) T-type current, has recently been cloned and characterized (11).
Comparison of the primary structures of the six identified vertebrate HVA ␣ 1 subtypes reveals a distinct separation between the L-type and non-L-type channel subfamilies. The similarity among subunits within each class is greater than that between the classes, suggesting that divergence between L-type and non-L-type channels constituted the first step in the evolution of known HVA Ca 2ϩ channels. This division between channel classes is maintained in the invertebrate Ca 2ϩ channels that have been cloned (for review see Ref. 12). Single homologues of both L-type and non-L-type channels have been found in Drosophila (13,14) and Aplysia (15), and other invertebrate Ca 2ϩ channel sequences (16 -18) exhibit clear resemblance to one of the two channel subfamilies. However, a thorough understanding of the relationship between invertebrate Ca 2ϩ channel structure and physiology requires functional expression of a cloned invertebrate Ca 2ϩ channel ␣ 1 subunit.
Cnidarians, which include jellyfish, anemones, and corals, are the earliest existing organisms to possess a neuromuscular system. Voltage-gated Ca 2ϩ currents have been recorded from neural and muscle cells of several cnidarian species (19 -21), including the scyphozoan jellyfish Cyanea capillata (22). We report here the cloning and functional expression of CyCa␣ 1 , a voltage-gated Ca 2ϩ channel ␣ 1 subunit from Cyanea. Although CyCa␣ 1 has the molecular structure of an L-type channel ␣ 1 subunit, certain aspects of the pharmacology and physiology of the expressed channel distinguish it from mammalian L-type channels.

EXPERIMENTAL PROCEDURES
Isolation of cDNA-RNA was extracted from Cyanea perirhopalial tissue (consisting of neurons, myoepithelial cells, endoderm, and mesoglea) by homogenization of the tissue in guanidinium isothiocyanate and centrifugation through CsCl. Two degenerate oligonucleotide primers were designed to correspond to highly conserved regions of Ca 2ϩ channel ␣ 1 subunit sequences. The sense primer (ATHACNATGGARG-GNTGG) corresponds to residues ITMEGW in the pore region of domain I, and the antisense primer (NCCNCCRAAIARYTGCAT, where I ϭ inosine) corresponds to residues MQLFGG in S5 of domain II. cDNA was synthesized from 10 g of total RNA in a reverse transcription (RT) reaction containing 50 pmol of antisense primer and 200 units of Su-perScript II RT (Life Technologies, Inc.). After a 2-h incubation at 42°C, the RT reaction was heat-inactivated, and 0.5 l of the cDNA was amplified in a PCR. The PCR, containing 50 pmol of the sense and antisense primers, 200 M dNTPs, and 2.5 units of Taq DNA polymerase (Fisher), incorporated a touchdown protocol, in which the annealing temperature was decreased from 55 to 45°C over 11 cycles, then was maintained at 45°C for 30 cycles. The resulting product was electrophoresed on an agarose gel and purified (Qiaex; Qiagen, Santa Clarita, CA), cloned into pGEM-T (Promega, Madison, WI), and sequenced using the Sequenase 2.0 version (Amersham Pharmacia Biotech) of the dideoxy method. The 5Ј end of the cDNA was generated by rapid amplification of cDNA ends (Ref. 23), in which a gene-specific antisense primer was used to prime cDNA synthesis from 1 g of total RNA. The cDNA was tailed with dCTP and terminal transferase (Promega) and then amplified in a PCR with a nested gene-specific primer and an oligo(dG/dI) primer. The 3Ј end of the open reading frame was determined by isolation of five overlapping fragments. Three of the fragments were generated by PCR from a Cyanea perirhopalial tissue cDNA library, using gene-specific and vector-specific primers, and further amplified with nested primers when necessary. The first fragment isolated in this manner terminated prematurely in a purine-rich region, so a second fragment was generated using inverse PCR, in which Cyanea genomic DNA was digested with HpaII and circularized by ligation, then amplified by PCR to extend the known sequence. Subsequent fragments were generated by PCR from the library and by rapid amplification of cDNA ends.
Once the sequences of all fragments of the cDNA were determined, Cyanea total RNA was used in separate RT reactions to create two additional independent cDNA pools. Overlapping fragments spanning the complete open reading frame of the cDNA were amplified from each cDNA pool and fully sequenced to establish a consensus sequence. A full-length cDNA clone corresponding to the consensus sequence was assembled in a vector (pXENEX1) constructed from portions of two other vectors, pAGA2 (Ref. 24; provided by Maninder Chopra, State University of New York, Buffalo) and pBScMXT (provided by Aguan Wei, Washington University). The multiple cloning site of pXENEX1 is flanked by the 5Ј-noncoding region of the RNA-4 gene from alfalfa mosaic virus and the 3Ј-noncoding region of the Xenopus ␤-globin gene. The incorporation of a consensus sequence for initiation of translation (25) required that the second amino acid of the open reading frame be changed from Phe to Val.
Northern Blot Analysis-100 g of total RNA from Cyanea perirhopalial tissue was electrophoresed on a glyoxal denaturing gel and blotted to positively charged nylon (Magnagraph; MSI, Westborough, MA). SP6 RNA polymerase was used to synthesize a 32 P-labeled antisense riboprobe from a cDNA clone encoding portions of domains I and II of CyCa␣ 1 . The blot was incubated with the probe under high stringency hybridization conditions (50% formamide, 5ϫ SSPE, 0.5% SDS, 60°C) and then was washed three times in 1ϫ SSPE, 0.5% SDS at 65°C, followed by one wash in 0.1ϫ SSPE, 0.5% SDS at 60°C.
Expression in Xenopus Oocytes-The CyCa␣ 1 construct was linearized with NgoMI, purified, and used to transcribe RNA using the T7 version of the mMessage mMachine in vitro transcription kit (Ambion, Austin, TX). Stage V-VI oocytes were removed from Xenopus under MS-222 anesthesia, defolliculated by treatment with 2 mg/ml collagenase in Ca 2ϩ -free ND96 medium (96 mM NaCl, 2 mM KCl, 1 mM MgCl 2 , and 5 mM HEPES, pH 7.4), and injected with 10 -50 ng of RNA, then incubated in sterile ND96 (containing 1.8 mM CaCl 2 , supplemented with 2.5 mM sodium pyruvate, 100 units/ml penicillin, 100 g/ml streptomycin, and 5% horse serum) at 17°C for 3-10 days. Oocytes were injected with 10 -20 nl 100 mM 1,2-bis(2-aminophenoxy)ethane-N,N,NЈ,NЈ-tetraacetic acid (BAPTA) (dissolved in 10 mM HEPES) at least 1 h prior to recording to eliminate artifacts caused by the endogenous Ca 2ϩ -activated chloride current (26). The standard bath solution consisted of 40 mM Sr(OH) 2 , 40 mM N-methylglucamine, 10 mM glucose, and 10 mM HEPES, adjusted to pH 7.4 with methanesulfonic acid (27). Where necessary, the Sr(OH) 2 in the bath solution was replaced by 40 mM Ca(OH) 2 or 40 mM Ba(OH) 2 . Most drugs were dissolved directly in Sr 2ϩ bath solution and tested by bath perfusion. Nifedipine and isradipine were dissolved in Me 2 SO and S(Ϫ)-BayK 8644 was dissolved in ethanol; then each was diluted in bath medium to a solvent concentration of 1% or less. Similar concentrations of Me 2 SO or ethanol had no significant effect on the current when administered alone. All DHPs were prepared and applied under subdued light.
Currents were recorded under two-electrode voltage clamp at room temperature (ϳ22°C). Electrodes were pulled from borosilicate glass and filled with 3 M KCl, resulting in an impedance of 0.1-1 M⍀. Oocytes were clamped at Ϫ90 mV (Ϫ110 mV in experiments testing inactivation), except as noted. Test protocols were administered using pClamp6 software. Leakage currents were subtracted on-line, using either a P/2 or P/4 protocol of hyperpolarizing prepulses.

RESULTS
Primary Structure of the Cyanea Ca 2ϩ Channel-The initial fragment of CyCa␣ 1 was isolated using two degenerate oligonucleotide primers corresponding to regions that are highly conserved in all known Ca 2ϩ channel ␣ 1 subunits. These primers were used with RT-PCR to amplify a cDNA segment from Cyanea perirhopalial tissue, and the resulting product was sequenced and found to resemble ␣ 1 subunit sequences. The fragment was extended using a series of PCR-based cloning techniques, leading to isolation of a 6857-base pair cDNA, which contains an open reading frame encoding a 1911-amino acid sequence (Fig. 1). The first methionine within the CyCa␣ 1 open reading frame, which is preceded by an in-frame stop codon 36 base pairs upstream, is not surrounded by a consensus sequence for initiation of translation (25), but this motif may not be strongly conserved in cnidarian messages (28). The 3Ј end of the sequence, which was amplified from an oligo(dT)primed cDNA library, contains several in-frame stop codons but no consensus polyadenylation signal upstream of the polyadenylated region, suggesting that this tail is spurious. In fact, Northern blot analysis of RNA from Cyanea perirhopalial tissue revealed a single band of approximately 8.5 kb (Fig. 2), indicating that another 1.5-2 kb of the untranslated region(s) has not yet been cloned.
The ␣ 1 subunit of Ca 2ϩ channels comprises four homologous domains arranged symmetrically to form the walls of a pore (6,29). Each domain consists of six transmembrane segments (S1-S6) and a loop between S5 and S6 that forms part of the ion selectivity pore of the channel. This four-domain structure is evident in CyCa␣ 1 . The sequence identity between CyCa␣ 1 and other HVA Ca 2ϩ channel ␣ 1 subunits (excluding the weakly conserved amino and carboxyl termini and the intracellular loop between domains II and III) ranges between 49 and 60%. In contrast, the corresponding regions of the LVA ␣ 1G subunit from rat (11) are only 34% identical to CyCa␣ 1 , and a voltagegated Na ϩ channel cloned from Cyanea (28) exhibits only 30% identity. The sequence of CyCa␣ 1 is more similar to those of mammalian L-type ␣ 1 subunits (59% average sequence identity in conserved regions) than to those of non-L-type ␣ 1 subunits (50% average sequence identity). However, the identity between CyCa␣ 1 and the L-type subunits ␣ 1C (60%), ␣ 1D (59%), and ␣ 1S (58%) does not identify CyCa␣ 1 as a homologue of any mammalian L-type subtype. Similarly, phylogenetic analysis of Ca 2ϩ channel ␣ 1 subunit sequences (Fig. 3) places CyCa␣ 1 among the HVA L-type ␣ 1 subunits, but suggests that CyCa␣ 1 appeared before the emergence of different subtypes within the L-type family.
Most of the highly conserved features of Ca 2ϩ channels appear in CyCa␣ 1 . Each pore-forming loop contains a glutamate residue (positions 339, 679, 1085, and 1372) that is present in all known voltage-gated Ca 2ϩ channels and is essential for establishing Ca 2ϩ selectivity (30,31). The S4 segments, which 10, GenBank TM accession number L15453). Sequences were aligned using ClustalW (63). Certain regions of the other sequences that exhibit little similarity to the Cyanea sequence are truncated to conserve space; the lengths of deleted regions are indicated in brackets. Residues in the Cyanea sequence that are identical to both of the L-type or both of the non-L-type channel sequences are shaded. The approximate position of transmembrane segments is based upon the alignment. The region involved in interaction with ␤ subunits (60) is labeled, and each of the glutamate residues found in the pore regions of all known voltage-gated Ca 2ϩ channel sequences is denoted by a diamond. The putative EF-hand region are involved in voltage sensitivity of Ca 2ϩ channels, display the conserved pattern of positively charged amino acids at every third position. The proximal region of the carboxyl terminus closely matches the consensus sequence for the EF-hand Ca 2ϩbinding motif (32,33) that plays a role in Ca 2ϩ -dependent inactivation of L-type channels (34). However, the region of the I-II interdomain loop that interacts with Ca 2ϩ channel ␤ subunits is not highly conserved in the CyCa␣ 1 protein. Of the nine residues found in all other vertebrate and invertebrate HVA ␣ 1 subunit sequences (except for one splice variant of Drosophila ␣ 1A (14)), only four (Gly-Tyr-Xaa-Xaa-Trp-Ile, beginning at residue 412) are found in CyCa␣ 1 . The Gln-Xaa-Xaa-Glu-Arg motif that has been shown in N-type channels to bind the G␤␥ sub-units of G proteins (35,36) is also absent from CyCa␣ 1 .
The protein contains six consensus cAMP-dependent protein kinase phosphorylation sites within the first two interdomain cytoplasmic loops and the carboxyl terminus. Twenty potential protein kinase C phosphorylation sites are distributed across all three interdomain loops and the amino and carboxyl termini, but no intracellular tyrosine kinase phosphorylation sites appear to be present in the protein. Three consensus N-linked glycosylation sites exist on regions of the protein thought to be exposed to the extracellular side of the membrane, two within the S5-S6 loop of domain I and one within the S1-S2 loop of domain IV.
Functional Characteristics of CyCa␣ 1 -Xenopus oocytes in- jected with 10 -50 ng of CyCa␣ 1 RNA exhibit a large inward current in response to depolarizing test pulses (Fig. 4A). The amplitude of the expressed current in 40 mM Ca 2ϩ or Sr 2ϩ (ϳ1-3 A) far exceeds that of the native Ca 2ϩ current in oocytes (Ͻ50 nA). This robust current was evident in the absence of exogenous ␤ or ␣ 2 ␦ subunits. The current activates within 2 ms, reaches a peak at 5-10 ms, and then inactivates. The threshold of activation of the current is approximately Ϫ20 mV, and maximal current occurs at ϩ15 to ϩ20 mV in 40 mM Sr 2ϩ (Fig. 4B). The value of h ϱ for half-maximal steady-state inactivation of the channel, based on a fit with a Boltzmann function, is Ϫ32 mV (Fig. 4C).
In contrast to most other expressed Ca 2ϩ channels, CyCa␣ 1 is less permeable to Ba 2ϩ than to Ca 2ϩ , and current is greatest in the presence of Sr 2ϩ (Fig. 5, A and B). In oocytes tested in the presence of 40 mM concentrations of each ion, the ratio of peak current levels for Sr 2ϩ :Ca 2ϩ :Ba 2ϩ was 1.00:0.73:0.65 (n ϭ 7). As in mammalian Ca 2ϩ channels (37)(38)(39), the I-V relationship for Ca 2ϩ permeation is shifted rightward (peak current at ϩ26 mV) compared with that for Ba 2ϩ or Sr 2ϩ (peak current at ϩ14 -15 mV), a shift that has been attributed to Ca 2ϩ -specific modulation of the gating of the ␣ 1 subunit (40). Consequently, the relative permeability of the channel to divalent cations varies with voltage, and the calculated ratio of maximal conductance for Sr 2ϩ :Ca 2ϩ :Ba 2ϩ is 1.00:0.82:0.66. In the presence of 40 mM Na ϩ and the absence of divalent cations, peak cur- The regions spanning IS5 to IIS6 and IIIS1 to the conserved region of the carboxyl tail were used for this analysis. Amino acid sequences were aligned using ClustalW. The aligned sequences were used to construct phylogenetic trees using the neighbor-joining method (64), as implemented in PHYLIP version 3.5 (65). The Cyanea sodium channel sequence (GenBank TM accession number L15445) was defined as the outgroup. The tree shown in the figure is identical in branching pattern to a tree produced using a maximum parsimony algorithm (PROTPARS). The branch lengths shown are proportional to calculated phylogenetic distance. Gen-Bank TM accession numbers are as follows (see Fig. 1  rents are about 12% those seen in the presence of Sr 2ϩ (data not shown). Due to the larger currents in the presence of Sr 2ϩ , all subsequent experiments were conducted with Sr 2ϩ as the charge carrier, except as noted.
As illustrated in Fig. 5A, the CyCa␣ 1 current inactivates in the presence of all three divalent cations. Although the rate of current decay varies slightly with voltage, Ba 2ϩ currents inactivate with a time constant between 30 and 36 ms at test potentials above ϩ10 mV, and Sr 2ϩ currents with a time constant between 27 and 30 ms (Fig. 5C). The rate of inactivation is increased when Ca 2ϩ is the permeant ion, indicating that CyCa␣ 1 exhibits the Ca 2ϩ -dependent inactivation characteristic of mammalian L-type Ca 2ϩ channels (34).
Pharmacology of CyCa␣ 1 -The CyCa␣ 1 current is surprisingly insensitive to DHPs. Whereas most mammalian L-type channels are fully blocked by 10 M nifedipine, only 50% inhibition of the CyCa␣ 1 current is produced by 100 M nifedipine (Fig. 6A). Similarly, the potent DHP isradipine produces only 36% inhibition at a concentration of 100 M. The potency of nifedipine or isradipine block is not enhanced by raising the holding potential from Ϫ90 mV to Ϫ60 mV or by coexpressing an exogenous Ca 2ϩ channel ␤ subunit (data not shown). CyCa␣ 1 is similarly insensitive to the phenylalkylamine verapamil; a 100 M concentration inhibits the current by 42%.
S(Ϫ)-BayK 8644, a DHP that acts as an agonist on vertebrate L-type channels, also has an unexpected action on the CyCa␣ 1 current. The current is not potentiated by concentra-  (n ϭ 7). The order in which bath solutions were applied was varied for each experiment, and the first bath solution tested was reapplied after all bath changes were complete to ensure that no current rundown had occurred. The data points were fitted by the Boltzmann function I ϭ g max (V Ϫ V rev )/(1 ϩ exp(Ϫ(V Ϫ V1 ⁄2 )/k), where I is whole cell current at test potential V; g max is maximal conductance; V rev is reversal potential; V1 ⁄2 is potential for half-activation; and k is slope factor. The parameter values for Sr 2ϩ , Ba 2ϩ , and Ca 2ϩ , respectively, are g max ϭ 101, 67, and 83 S; V rev ϭ 69, 67, and 74 mV; V1 ⁄2 ϭ 2.4, 0.6, and 12.8 mV; and k ϭ 6.7, 7.5, and 7.6 mV. Data points reflect peak current elicited by a 250-ms voltage step to ϩ20 mV from a holding potential of Ϫ90 mV. B, Block of the CyCa␣ 1 Sr 2ϩ current by Cd 2ϩ (circles; n ϭ 6) or Ni 2ϩ (squares; n ϭ 6). Recording conditions are as described in A.
The N-type channel blocker -conotoxin GVIA (10 M) and the P/Q-type channel ligands -conotoxin MVIIC (5 M) and -agatoxin IVA (1 M) have no effect on the CyCa␣ 1 current (data not shown). In addition, CyCa␣ 1 is potently blocked by Cd 2ϩ (IC 50 ϭ 2.1 M; Fig. 6B), but, in contrast to R-type or T-type channels, is not highly sensitive to Ni 2ϩ block (IC 50 ϭ 570 M).

CyCa␣ 1 Represents an Early L-type Ca 2ϩ
Channel-Cnidarians, the earliest existing metazoans, appeared 600 million years before the emergence of mammals (41). Voltage-gated Ca 2ϩ currents have been identified in more ancient eukaryotes, such as protists (42)(43)(44) and plants (45), but cnidarians are the simplest organisms in which movement is governed by a neuromuscular system. Although the tissue distribution of CyCa␣ 1 is unknown, the channel is likely to be a component of Cyanea neural or muscle cells. The perirhopalial tissue from which the CyCa␣ 1 cDNA was derived contains several cell types, including neurons, muscle, ectodermal and endodermal epithelial cells, and cnidocytes. Ca 2ϩ currents in Cyanea muscle cells have not yet been characterized, but Cyanea neurons possess a high voltage-activated Ca 2ϩ current of undetermined type (22). Scyphozoan jellyfish do not have electrically excitable epithelial cells, however, and no inward currents can be recorded from their cnidocytes (46). This evidence suggests that CyCa␣ 1 represents one of the earliest examples of a neuromuscular voltage-gated Ca 2ϩ channel.
Sequence analysis divides all previously known HVA Ca 2ϩ channel ␣ 1 subunits into two classes, termed L-type and non-L-type. Although the CyCa␣ 1 sequence varies significantly from those of mammalian channels, it clearly shares more amino acid identity with L-type channels than with non-L-type channels. An ␣ 1 subunit that appeared before the divergence of the two HVA channel subfamilies would be expected to resemble members of both classes equally, as is seen with the LVA ␣ 1G subunit. The identification of CyCa␣ 1 as an L-type channel implies that the separation of HVA Ca 2ϩ channels into L-type and non-L-type classes occurred prior to or concurrent with the evolution of neuromuscular systems. Non-L-type channel genes have been identified in nematodes (17) and platyhelminths, 2 but we have not yet obtained molecular evidence for the existence of non-L-type Ca 2ϩ channels in Cyanea.
CyCa␣ 1 does not preferentially resemble any one of the mammalian subtypes (␣ 1S , ␣ 1C , and ␣ 1D ) of the L-type class. Instead, phylogenetic analysis indicates that CyCa␣ 1 , and perhaps other invertebrate L-type channels, represents an ancestral form that emerged before the appearance of the different mammalian L-type subtypes. Therefore, CyCa␣ 1 has not been assigned a subscript to conform to the accepted nomenclature for vertebrate Ca 2ϩ channels. Although subtype designations have been assigned to certain invertebrate Ca 2ϩ channels on the basis of sequence similarity, multiple molecular subtypes within the L-type or non-L-type class have thus far been identified only in vertebrates (e.g. Ref. 47).
The primary structure of CyCa␣ 1 also relates to evolution within the voltage-gated ion channel superfamily. The poreforming subunit of voltage-gated Na ϩ channels, which has a structure homologous to that of the Ca 2ϩ channel ␣ 1 subunit, is thought to have arisen from Ca 2ϩ channels at a time roughly coinciding with the appearance of metazoans (48). However, CyCa␣ 1 exhibits no greater similarity to a Cyanea Na ϩ channel (28) than do mammalian Ca 2ϩ channel ␣ 1 subunits; thus, Na ϩ channels and Ca 2ϩ channels are already highly divergent at the level of cnidarians.
The CyCa␣ 1 Current Differs from Vertebrate L-type Currents-The molecular structure of CyCa␣ 1 clearly identifies it as L-type, but the pharmacological response of the expressed current differs considerably from that of mammalian L-type currents. Although CyCa␣ 1 is unaffected by peptide toxins specific for non-L-type channels and is only weakly sensitive to Ni 2ϩ , which potently blocks T-type and R-type channels, the CyCa␣ 1 current is also quite insensitive to block by DHPs. Significant current remains in the presence of 10 M nifedipine, a concentration sufficient to completely block the rabbit ␣ 1C channel expressed in oocytes (3), and full inhibition of the CyCa␣ 1 current is not evident below 1 mM nifedipine. The DHP isradipine and the phenylalkylamine verapamil are similarly ineffective in blocking the current. Furthermore, S(Ϫ)-BayK 8644, a DHP enantiomer that potentiates vertebrate L-type currents, produces only weak inhibition of the CyCa␣ 1 current. In fact, the low potency of the antagonist and agonist actions, as well as the shallowness of the concentration-response curves, indicates that the moderate inhibition seen at higher drug concentrations may be due to nonspecific effects.
The insensitivity of CyCa␣ 1 to DHPs implies that the structure of the DHP-binding pocket identified in higher L-type channels is altered in CyCa␣ 1 . The residues throughout domains III and IV that have been implicated in conferring DHP sensitivity to L-type channels (49 -53) are identical between CyCa␣ 1 and ␣ 1C in all but two positions. The substitution of Gly-903 in CyCa␣ 1 in place of Phe may not be critical; an analogous mutation in the human ␣ 1C channel has little effect on DHP block (49). More significant is the presence of Ile-1128. The Met residue at the corresponding position in the ␣ 1C subunit influences both block and potentiation by DHPs (52)(53)(54). The Musca and Drosophila L-type ␣ 1 subunits also contain a hydrophobic substitution for Met at this site; in fact, the replacement of Met-1188 (␣ 1C numbering) is responsible for the decreased response of Musca-␣ 1C chimeric channels to DHPs (52). The experiments presented herein directly show DHP insensitivity in an invertebrate L-type ␣ 1 subunit, which is in accord with reports that Ca 2ϩ currents in lower invertebrates exhibit decreased sensitivity to DHP block. The Ca 2ϩ current in Paramecium is unaffected by DHPs (43), and neuronal Ca 2ϩ currents in the hydrozoan jellyfish Polyorchis (21) and the flatworm Bdelloura (55) are only partially blocked by 10 -100 M nifedipine (although these Ca 2ϩ currents may not be gated by L-type channel homologues). Molluscs are the earliest organisms in which potent block of Ca 2ϩ currents by DHPs has been reported (56), although the current is unaffected by racemic BayK 8644. The present results demonstrate that the actions of DHPs, which are generally used to define a current as L-type, may not reliably identify L-type channel homologues in lower organisms.
The CyCa␣ 1 subunit also exhibits a distinctive pattern of whole-cell permeability to divalent cations. Replacement of Ca 2ϩ by Ba 2ϩ results in decreased current in oocytes expressing CyCa␣ 1 , and Sr 2ϩ produces larger currents than Ca 2ϩ . This selectivity profile (I Sr Ͼ I Ca Ͼ I Ba ) is distinct from that of most expressed vertebrate ␣ 1 subunits (I Ba Ͼ I Sr Ͼ I Ca ). However, the permeability of the non-L-type rat ␣ 1E subunit in oocytes (39), which differs even from that of other mammalian ␣ 1E subunits (38), is similar to that of CyCa␣ 1 . Inferring channel permeability from whole-cell recordings is made difficult by effects of divalent cations on the mean open time of the channel (57), and patch clamp experiments are necessary to accurately determine permeability. Nevertheless, the similar whole-cell profiles of the Cyanea and rat ␣ 1E channels indicate that they may share structural features distinct from those of most known vertebrate HVA Ca 2ϩ channels. Alignment of the pore region sequences of CyCa␣ 1 and mammalian ␣ 1 subunits reveals no single amino acid substitution unique to CyCa␣ 1 and rat ␣ 1E . An aspartate residue found in the domain I pore region of most ␣ 1 subunits is replaced in rat ␣ 1E by a threonine (Thr-264), which has been proposed to play a role in altered ionic selectivity (39). However, the aspartate is conserved in CyCa␣ 1 (Asp-343).
One region of CyCa␣ 1 that is poorly conserved is the domain in the I-II intracellular loop that interacts with Ca 2ϩ channel ␤ subunits. The ␤ subunit alters the activation and inactivation kinetics of the ␣ 1 subunit and increases current amplitude (5,58), and is also involved in trafficking of the ␣ 1 subunit to the plasma membrane (59). The I-II loop of the ␣ 1 subunit contains an 18-amino acid cassette to which the ␤ subunit binds (60,61). Nine residues are conserved within this cassette in all known vertebrate Ca 2ϩ channel sequences (QQXEXXLXGYX-XWIXXXE), but only four are found in CyCa␣ 1 . Of these four, three (Tyr-413, Trp-416, and Ile-417) have been shown to be the most critical residues for ␤ subunit binding (62). Although CyCa␣ 1 does not require an exogenous ␤ subunit for functional expression, we have cloned a ␤ subunit from Cyanea that is able to modulate the activity of CyCa␣ 1 . 3 These findings illustrate the value of studying a phylogenetically distant voltage-gated Ca 2ϩ channel. CyCa␣ 1 diverged from the ancestor of vertebrate Ca 2ϩ channels over half a billion years ago, creating the opportunity for variation in all but the most structurally constrained regions of the protein.
The ability to correlate this novel structural information with the physiology of the expressed current provides a new avenue for examining Ca 2ϩ channel function.