Transient and Big Are Key Features of an Invertebrate T-type Channel (LCav3) from the Central Nervous System of Lymnaea stagnalis*

Here we describe features of the first non-mammalian T-type calcium channel (LCav3) expressed in vitro. This molluscan channel possesses combined biophysical properties that are reminiscent of all mammalian T-type channels. It exhibits T-type features such as “transient” kinetics, but the “tiny” label, usually associated with Ba2+ conductance, is hard to reconcile with the “bigness” of this channel in many respects. LCav3 is 25% larger than any voltage-gated ion channel expressed to date. It codes for a massive, 322-kDa protein that conducts large macroscopic currents in vitro. LCav3 is also the most abundant Ca2+ channel transcript in the snail nervous system. A window current at typical resting potentials appears to be at least as large as that reported for mammalian channels. This distant gene provides a unique perspective to analyze the structural, functional, drug binding, and evolutionary aspects of T-type channels.

T-type calcium channels open in response to slight depolarizations in the low voltage range. Paradoxically, they are also recruited after membrane hyperpolarization as occurs during rebound burst firing (1). A window current of T-type channels is a feature that permits Ca 2ϩ entry at rest (2) and contributes to differentiation and growth promoting functions in both excitable and non-excitable cells (3). T-type channels are also a leading pharmaceutical drug target and are implicated in a wide range of conditions such as epilepsy, pain, hypertension, cancer, and mental disorders (4).
T-type Ca 2ϩ currents were first measured in starfish eggs using a two-electrode voltage clamp (5). Currents conducted by "Channel I" were evoked by small depolarizations (low voltageactivated), visible as a small hump in a current amplitude versus test potential plot, appearing inconsequential beside the Channel II currents elicited by larger depolarizations (high voltageactivated). Ca 2ϩ channel types would be discriminated further by Tsien and co-workers (6) on the basis of properties where Ba 2ϩ is the charge carrier. High voltage-activated L-type channels have a large unitary Ba 2ϩ conductance with long-lasting openings, N-type (or non-L-type) channels are typically associated with neurons of intermediate unitary conductance, and the low voltage-activated, T-type channels produce transient currents that are of tiny unitary conductance in Ba 2ϩ and close slowly upon membrane repolarization, producing a slowly deactivating tail current (6).
T-type channels remain as the least understood among the Ca 2ϩ channel families. Although most of the 10 mammalian Ca 2ϩ channel genes were characterized in the late 1980s, an additional decade was required for a description of the three T-type genes, Ca v 3.1 (␣1G), Ca v 3.2 (␣1H), and Ca v 3.3 (␣1I) (7). Progress in understanding T-type channel functions continues to be hampered by the lack of highly selective blockers that discriminate between Ca v 3 channel types or separate Ca v 3 channels from related L-type (Ca v 1) and non-L-type (Ca v 2) Ca 2ϩ channels, which usually produce more robust Ca 2ϩ entry into the same cells (7).
Here we describe the in vitro expression characteristics of the first non-mammalian, T-type channel, LCa v 3, cloned from the pond snail, Lymnaea stagnalis. This structurally distant channel has quintessential features of T-types such as transient kinetics. LCa v 3 is big in many respects, such as its protein size; it expresses large macroscopic currents in human cells, it is the most abundant Ca 2ϩ channel transcript in the snail nervous system, and it generates window currents that appear to be at least as large as those reported for mammalian channels.
LCa v 3 provides a unique perspective to analyze the structure, function, and drug binding of T-type channels and serves as a useful surrogate in residue swapping experiments. Searches for the fundamental mechanisms that regulate this singleton invertebrate T-type channel will be facilitated by the simple molluscan preparation, where accessible and identified neurons underlying well described behaviors can be studied in isolated Lymnaea neurons, cultured synapses, or within intact, identified networks in situ. Also, LCa v 3 provides nourishment for evolutionary speculation. Although the first gastropods (500 million years ago) are likely quite distant from this ancestral branch point, the extant snail homolog, LCa v 3, is reminiscent of the gene that predates the speciation that led to the emergence of the three distinct, mammalian T-type channel genes.

EXPERIMENTAL PROCEDURES
Cloning and Sequencing of LCa v 3-The complete open reading frame for LCa v 3 was determined from at least three independent, overlapping DNA fragments from the PCR screening of L. stagnalis central nervous system ZAP cDNA libraries to generate a consensus gene. The full-length 9031-bp cDNA transcript is available in DDBJ/EMBL/ GenBank TM databases under accession no. AF484084 and replaced a previous partial coding sequence entry of 5991 bp. The final clone was assembled from four overlapping PCR with sticky ends (numbered by cDNA transcript positions), XhoI-SpeI (209 -2865), SpeI-SalI (2812-4544), SalI-MluI (4503-6874), and MluI-BamHI (6850 -8869). Silent mutations were created in a Kozak consensus sequence upstream of the start codon (209 -211), an MluI site (6858 -6863), and several hairpin structures thought to interfere with the site-directed mutagenesis reaction (5225-5285). The full-length LCa v 3 coding sequence was assembled between XhoI and BamHI sites in bicistronic vector pIRES2-EGFP 2 (Clontech). Low frequency of positive recombinants during cloning and the slow rate of growth of the full-length plasmid in bacteria (five full days before a colony appears on a bacterial plate after transformation) suggest that the plasmid insert is toxic to bacteria.
Transfections-HEK-293T cells (M. Calos, Stanford University) were cultured in Dulbecco's modified Eagle's medium (Sigma) with 10% fetal bovine serum (Sigma) and supplemented with 0.5% (v/v) penicillin-streptomycin solution (Sigma). For electrophysiology, 6 g of the LCa v 3 pIRES2-EGFP construct was transfected into cells at 40 -50% confluency using the standard Ca 2ϩ phosphate transfection method. After overnight transfection, the cells were washed twice with culture media and incubated at 28°C in a humidified, 5% CO 2 chamber for 3 days. After incubation, cells were detached using a trypsin-EDTA solution (Sigma), plated at 10% confluency onto glass coverslips, and incubated at 37°C for 4 h because adhesion to the glass substrate requires warmer temperatures (8).
Whole Cell Patch Clamp Recordings-Whole-cell recordings were carried out at 23°C using either a 5 mM external Ca 2ϩ solution (5 mM CaCl 2 , 166 mM tetraethylammonium chloride, 10 mM HEPES, pH 7.4) or 5 mM external Ba 2ϩ solution (5 mM BaCl 2 , 166 mM tetraethylammonium chloride, 10 mM HEPES pH 7.4) and an internal solution consisting of 125 mM CsCl, 10 mM EGTA, 2 mM CaCl 2 , 1 mM MgCl 2 , 4 mM MgATP, 0.3 mM Tris-GTP, and 10 mM HEPES, pH 7.2. Recordings were obtained using an Axopatch 200B amplifier, sampled to a PC through a Digidata 1440a A/D converter. Data were filtered at 2 kHz and digitized at 5 kHz and acquired using pCLAMP 10.1 software (Molecular Devices). The pipette resistance was maintained between 3 and 5 megaohms, and the typical access resistance was between 4 and 5 megaohms. Only recordings with minimal leak (Ͻ10%) and small current sizes (Ͻ2 nA) were used for analysis, and offline leak subtraction was carried out using the Clampfit 10.1 software (Molecular Devices). Series resistance was compensated to 70% (prediction and correction; 10-s lag). A gravity flow system was used to perfuse 5 mM Ca 2ϩ -or Ba 2ϩ -containing extracellular solution or 5 mM Ca 2ϩ external solution containing solubilized Ni 2ϩ (Sigma) or mibefradil (Sigma).
Data Analysis-Ca 2ϩ current activation curves were constructed by converting the peak current values from each cur-rent-voltage relationship data set to conductance using the equation gCa ϭ I peak /(V command Ϫ E Ca ), where I peak is the peak current, V command is the command pulse potential, gCa is the calcium conductane, and E Ca is the Ca 2ϩ reversal potential as determined by linear extrapolation of the current values in the ascending portion of the current-voltage relationships. Conductance values were then normalized and individually fitted with the Boltzmann equation, g/g max ϭ (1 ϩ (exp(ϪV command Ϫ V1 ⁄ 2 )/K)) Ϫ 1, where g is the peak conductance, g max is the maximal peak Ca 2ϩ conductance, V command is the conditioning potential, V1 ⁄ 2 is the half-maximal activation, and k is the activation slope factor. The steady-state inactivation curves were constructed by plotting normalized current (peak test pulse current/peak prepulse current) as a function of the inactivating potential. The data were fitted with a Boltzmann equation, where I is the peak test pulse current, I max is the peak test pulse current when the conditioning pulse was Ϫ110 mV, V inact and V1 ⁄ 2 are the conditioning potential and the half-maximal inactivation, respectively, and k is the inactivation slope factor. Kinetics of activation, inactivation, and deactivation were determined by fitting monoexponential functions over the growing or decaying phases of each current trace using the software Clampfit 10.1.
Antibody Production-LCa v 3 I-II linker coding sequence (1976 -2575 bp) was PCR-amplified and cloned into the bacterial protein expression vector pET-22b(ϩ) (Novagen) via NdeI and XhoI restriction sites. Peptide expression was induced by 1 mM isopropyl 1-thio-␤-D-galactopyranoside in Rosetta TM (DE3) cells (Novagen) transformed with the pET22b(ϩ) plasmid-containing construct. Supernatant of lysed bacterial cells containing His 6 LCa v 3 I-II linker expression was run and washed through a column containing Ni 2ϩcharged His⅐Bind resin (Novagen), then eluted off of the beads, dialyzed, and quantified using the Bradford assay. A rabbit was injected three times with recombinant proteins emulsified with Freund's complete adjuvant for the first injection and Freund's incomplete adjuvant (Sigma) for the subsequent injections. IgG rabbit antiserum was tested for immune reactivity with the antigen by Western blotting.
Immunolabeling of Recombinant LCa v 3 in HEK-293T Cells-HEK-293T cells were transfected with either 8 g of LCa v 3 in pIRES2-EGFP alone or with 3.2 g of LCa v 1 ␣1 subunit in pIRES2-EGFP (9) plus 2.4 g of the rat ␤1 subunit in pMT2 and 2.4 g of rat ␣ 2 ␦ in pMT2. Cells were washed after transfection and incubated at 28°C for 1 week before trypsinization and plating onto glass coverslips. Cells were then fixed with 1% paraformaldehyde in PBS overnight at 4°C, washed twice with PBS, then permeabilized using phosphate-buffered saline containing 0.2% Tween 20 (PBS-T) for 10 min at room temperature. 1:500 primary antibody or preimmune serum was applied to preblocked cells overnight in PBS-T containing 3% bovine serum albumin. 3ϫ-washed cells were incubated with 1:1000 diluted AlexaFluor 594 goat anti-rabbit secondary antibody in PBS-T containing 3% bovine serum albumin for 1 h at 23°C, washed 4ϫ, and imaged at 40ϫ magnification with a Zeiss AxioObserver Z1 inverted epifluorescent microscope to detect AlexaFluor 594 antibody and EGFP. Images were captured using Zeiss AxioVision software, an brightness/contrast was adjusted using Adobe Photoshop.
Southern Blot-15-g aliquots of genomic DNA isolated from Lymnaea tissue were digested with EcoRV, HindIII, EcoRI, and XhoI, and DNA fragments were separated through a 1% agarose gel. Digested DNA was transferred onto a positively charged nylon membrane (Roche Applied Science), and a standard hybridization procedure was carried out using manufacturer's instructions (EasyHyb, Roche Applied Science). Membranes were probed with a gel-purified 597-bp PCR product of the LCa v 3 gene (1979 -2575 bp) incorporated with DIG-11-dUTP (Roche Applied Science). The probe was localized on the membrane using anti-digoxigenin alkaline phosphatase-conjugated antibody (Roche Applied Science; 1:5000 dilution) and color substrate solution nitro blue tetrazolium/5-bromo-4-chloro-3-indolyl phosphate (Roche Applied Science).
Semiquantitative RT-PCR-RNA was extracted from the central nervous system of L. stagnalis using Tri Reagent (Sigma) and optimized from standard methods (10). RNA was treated with DNase (Fermentas) to remove contaminant DNA for RT-PCR analysis. RNA was quantified by spectrophotometry and visualized by gel electrophoresis to confirm the lack of degradation. First-strand cDNA were synthesized from RNA at 54°C for 80 min and 70°C for 15 min using either oligo-dT or random hexamer primers and Superscript III reverse transcriptase (Invitrogen) and controls were prepared that lacked reverse transcriptase. Primers used for cDNA amplification were designed to have similar melting temperatures, minimal secondary structure, and amplified fragments of similar sizes (500 -600 bp). PCR primers spanned sequences from the following genes (GenBank TM accession nos.  It is coded by a 9031-bp cDNA transcript that forms a 2886-amino acid protein with a molecular mass of 322 kDa. A, the N terminus closely matches with a putative start site derived from marine snail A. californica EST (EB302921) and slightly resembles the N and C termini of human Ca v 3.1-3.3. B, LCa v 3 is 1.25ϫ larger than human Ca v 3 channels and 1.5ϫ larger than nematode T-type, cca-1B, and all other four repeat ion channels. C, LCa v 3 is larger than human Ca v 3 channels in the N and C terminus and also the I-II and II-III cytoplasmic linkers.

Identity of an Invertebrate T-type Channel (LCa v 3)-A novel
invertebrate T-type channel transcript (9031 bp) was assembled from cDNA derived from the central nervous system of the freshwater pond snail, L. stagnalis, with a coding region that starts as an almost perfect match to an Expressed Sequence Tag data base entry from the marine snail Aplysia californica (accession no. EB302921). The LCa v 3 open reading frame predicts a 2886-amino acid protein, with an estimated ϳ322-kDa molecular mass. The start and end of the LCa v 3 amino acid sequence also resemble those of the human Ca v 3.1 channel (Fig.  1A). Consistent with other related cation channels of this type, LCa v 3 has four repeat domains (DI to DIV) with each domain containing six membrane-spanning segments similar to the voltage-gated K channels (Fig. 3A). LCa v 3 is the largest voltagegated ion channel expressed to date, being 25% larger than the mammalian T-types, 50% larger than a T-type homologue from Caenorhabditis elegans (cca-1b), and significantly larger than other Ca 2ϩ channels (Ca v 1, Ca v 2), sodium channels (Na v ), and NALCN (Fig. 1B). The extra size is mostly due to long cytoplasmic N and C termini and the cytoplasmic I-II and II-III linkers (Fig. 1C).
Snails, like most invertebrates, appear to have branched before the genomic duplication events that led to the expansion of gene isoforms and as such bear only single representatives for each of the three Ca 2ϩ channel gene families (Ca v 1, Ca v 2, and Ca v 3) compared with the 10 different Ca 2ϩ channel genes in mammals ( Fig. 2A). The snail T-type gene diverges dramatically in amino acid sequence from the mammalian genes (mean identity/similarity ϭ 37.6/46.1), whereas the mammalian homologues are clustered closer together (mean identity/similarity ϭ 50.7/57.9) (Fig. 2C). The single copy nature of the LCa v 3 gene in the genome is evidenced by Southern blotting, producing a single banding hybridization pattern, except for a gel lane where the restriction enzyme (EcoRV) cuts the genomic DNA within the probe sequence, leading to two bands of weaker intensity (lane a, Fig. 2B). The LCa v 3 transcript is more abundant than other Ca 2ϩ channels (LCa v 1 and LCa v 2) or LNALCN in central nervous system tissue as measured by semiquantitative RT-PCR (Fig.  2D). The higher expression of LCa v 3 compared with these others was also found by single cell, quantitative real-time PCR of individual VD4 neurons in a previous study using different primer sets (see Spafford et al. (11) and Fig. 2). Fig. 3A illustrates a running window of similarity between aligned LCa v 3 and human Ca v 3.3 channel protein sequences. The strongest homology is observed in the six-transmembrane segments and pore (P)-loops of each domain (Fig. 3A). Side chains of conserved negative residues lining the four P-loops contribute to a DDEE selectivity filter in T-type channels (7), including LCa v 3 (Fig. 3B). The pores of LCa v 3 and all voltagegated Ca 2ϩ channels also include a highly conserved aspartate adjacent to the selectivity-filter glutamate in Domain II (Fig.  3B), which may serve to attract incoming Ca 2ϩ ions to the ion selective pore. 3 One noticeably conserved region outside of the membrane-spanning domains is the gating brake present in the proximal I-II cytoplasmic linker (Fig. 3C). Comparison of LCa v 3 with Cav3.3 residues suggests conserved elements in the gating brake, including a putative helix-loop-helix hydrophobic 3 B. Zhorov, personal communication. A T-type probe hybridized to create a banding pattern (white arrows) on the blot was created from membrane transfer of genomic DNA digested with either EcoRV(a), HindIII (b), EcoRI (c), or XhoI (d). The probe contained an EcoRV restriction site, so the probe hybridized to two genomic DNA fragments digested with EcoRV. D, densitometric intensity of RT-PCR bands (illustrated in the inset) was generated from Lymnaea brain tissue. core, a putative salt bridge, and potential protein-protein interaction sites facing away from the hydrophobic core (Fig. 3C). 4 Expression Characteristics of LCa v 3 in HEK-293T Cells-Transient transfection of LCa v 3 cDNA contained in pIRES2-EGFP vector reveals membrane-delimited staining of HEK-293T cells (box, Fig. 4A) with a rabbit polyclonal antibody generated against the I-II cytoplasmic linker of LCa v 3 (see Fig. 3A for the relative location of the epitope). Antibody staining was not apparent in LCa v 1-transfected cells (9) or when preimmune serum was used to detect LCa v 3 (Fig. 4A). Channel expression levels generally correspond to the EGFP intensity in HEK-293T cells, as would be expected with the LCa v 3 cDNA expressed on the same mRNA as EGFP using the bicistronic expression vector, pIRES2-EGFP. The optimal level of HEK-293T expression for electrophysiological recording (200 pA to 1.5 nA) corresponds to 3 days after transfection, whereas larger currents of up to 10 nA were possible by allowing protein expression to continue for up to 6 days (Fig. 4B). Typical transient kinetics of T-type currents are revealed in whole cell recordings of LCa v 3. Small voltage steps above threshold (Ϫ70 to Ϫ65 mV) are slow to activate and inactivate (requiring 10s of ms). Larger voltage steps elicit currents with progressively faster activation and inactivation kinetics that cross over each other with each successive step toward a maximal current for a voltage step of Ϫ35 mV (Fig. 4C). of inactivation kinetics follows the change in the time to peak and of activation kinetics with increasing voltage steps (Fig. 4E) (see Table 1 for a detailed comparison of biophysical parameters between Ca v 3 channels).
The pairing of activation and inactivation with LCa v 3 is typical of T-type currents and has led some to suggest that T-type channel inactivation is voltage-independent (12). The voltage sensitivity of LCa v 3 approximates a typical low threshold T-type current that peaks at Ϫ35 mV (Fig. 4d), although technically LCa v 3 is slightly lower threshold than mammalian T-types (peak between Ϫ30 and Ϫ25 mV). Channel availability at steady state was assessed after a 1-s prepulse protocol (Fig.  5A) and a 5-s prepulse protocol (data not shown), revealing a surprisingly steep and positively shifted availability curve compared with mammalian T-type channels (Fig. 5B). The combination of the very low threshold of channel activation and the 4 E. Perez-Reyes, personal communication.  (S1-S4). Illustrated is the position in the I-II linker where LCa v 3 polyclonal antibody (Ab) was generated in rabbits against a 200-amino acid peptide. B, shown is amino acid sequence alignment of the re-entrant P-loop located between S5 and S6 of each of the four domains illustrating the signature sequence (EEDD locus) that influences Ca 2ϩ ion permeation and selectivity. The conserved aspartate residue (1097 in LCa v 3) in a position downstream of the selectivity filter glutamate residue is positioned to attract incoming Ca 2ϩ ions to the pore. 3 LCa v 3 contains a neutral isoleucine in the outer pore at position 468 where mammalian T-type channels have a negatively charged residue (Glu or Asp) that influence pore blocking drugs. C, alignment of the cytoplasmic gating brake in proximal I-II linker is shown. The gating brake is thought to prevent T-type channel gating at more hyperpolarized potentials. MARCH 5, 2010 • VOLUME 285 • NUMBER 10 large fraction of possible available channels creates a potentially large and persistent window current near the resting membrane potential of typical neurons (2) (Fig. 5C). An estimated window current was gathered by the product of available channels under steady-state conditions and the relative peak conductance. As many as 1.8% of total T-type channels may contribute to this current at Ϫ65 mV (Fig. 5C), a value at least as high as that calculated for recombinant mammalian channels gathered under similar conditions. A measure of the persistent, steadystate current amplitude was assessed after 1 s of sustained potentials held from a range of Ϫ70 to Ϫ50 mV in 5-mV increments. The largest, persistent current corresponded to the estimated maximal window current size at resting membrane potentials of Ϫ65 mV (Fig. 5c, inset).

Features of a Novel Invertebrate T-type Channel
One of the characteristic features of T-type channels is a slow rate of deactivation (7). Deactivation is measured as the rate of current decay from a tail current generated by the rapid return to lower, more hyperpolarized potentials with maximally opened channels (held at Ϫ35 mV) (Fig. 6). Deactivation rates of LCa v 3 are fastest at hyperpolarized potentials (Ϫ110 mV) and quickly slow with depolarization steps to resting potentials (Ϫ60 mV) (Fig. 6). LCa v 3 fits within the faster end of the range of deactivation kinetics for mammalian T-type channels but is still manyfold slower than Ca v 1 and Ca v 2 channels. The slowness of deactivation kinetics suggests that native LCa v 3 currents may pass a deactivating tail current upon membrane repolarization.
Macroscopic, native Ca 2ϩ currents are typically equal or larger than Ba 2ϩ currents at equimolar concentrations, although the unitary conductance is reported to be equal in high Ca 2ϩ or Ba 2ϩ (13). Macroscopic Ca 2ϩ currents range from smaller, equal, or larger than Ba 2ϩ currents for recombinant Ca v 3.2, Ca v 3.3, or Ca v 3.1 channels, respectively (14). Reasons for the relative differences in Ca 2ϩ and Ba 2ϩ permeability of different channel types are not clearly understood. LCa v 3 resembles Ca v 3.2 and other high voltage-activated snail Ca 2ϩ channels, conducting larger amplitude whole cell Ba 2ϩ currents than Ca 2ϩ currents (Fig. 7A) (7). Ba 2ϩ as a charge carrier results in a slight hyperpolarizing shift in the current-voltage relationships compared with Ca 2ϩ (Fig. 7B), but there is still an ϳ50% increase in whole cell Ba 2ϩ conductance compared with Ca 2ϩ in the absence of driving force changes (Fig. 7C). Kinetics are also faster when Ba 2ϩ is the charge carrier, with faster time to

. Transient transfection of HEK-293T cells harboring the pIRES2-EGFP plasmid containing invertebrate T-type channel cDNA reveal highly abundant channels and characteristic T-type channel properties. A, membrane delimited staining of LCa v 3 (inset) is evident in EGFP-positive cells but only with
LCa v 3specific antibody and not with preimmune serum or with LCa v 1-transfected cells. B, the box chart indicates the current density (pA/pF (picofarads)) of LCa v 3 expression on 3 or 6 days after transfection. The box chart also illustrates mean, median Ϯ 1 S.D., min/max current densities. C, sample LCa v 3 currents are shown in response to 5-ms voltage steps from a Ϫ110-mV holding potential. Illustrated is an ensemble of rapidly activating and inactivating Ca 2ϩ currents where each trace "crosses over" the previous one from rest to peak, and the resulting normalized peak currents are plotted as a function of voltage step, indicating low threshold of activation (Ϫ65 mV) and maximal currents generated at a step to Ϫ35 mV (D). Current-voltage relationships were curve-fitted with an Ohmic-Boltzmann function. E, the increase in inactivation kinetics ( inact ) closely follows the increasing speed at which the current approaches peak (t peak ), also reflected in the faster rate of activation, curve-fitted and represented by act . peak (Fig. 7D) associated with more rapid inactivation kinetics (Fig. 7E). Ca 2ϩ -dependent inactivation typically associated with Ca v 1 channels is not a property of LCa v 3 or other T-type channels (14).
Ni 2ϩ traditionally has been considered to be a blocker that distinguishes T-types from other channels, but only one of the three cloned mammalian T-channels, Ca v 3.2, is strongly inhibited by Ni 2ϩ (15). LCa v 3 is approximately equally sensitive to Ni 2ϩ as Ca v 3.1 and Ca v 3.3, with an IC 50 of 300 Ϯ 29.24 M (Fig.  8A), but all of these T-type channels are ϳ20 -60-fold less sensitive than Ca v 3.2 (15). Lee and co-workers (16) identified that the unusual Ni 2ϩ sensitivity of Ca v 3.2 critically involves His-191, imbedded in a helix-turn-helix motif known as the S3b-S4 voltage sensor paddle in Domain I (17) (Fig. 8C). Interestingly, LCa v 3 does not have a corresponding His-191 residue of Ca v 3.2, but neither does the sequence of the S3b-S4 voltage sensor paddle of LCa v 3 compare well with any of the Ca v 3 channels (Fig.  8C). LCa v 3 has an eight-amino acid insert in this short linker region and extra positive and negative charges compared with the mammalian Ca v 3 channels (Fig. 8C). The Ni 2ϩ dose-response curve does not perfectly fit the data (Fig. 8B), but a biphasic dose-response curve does, having a high affinity IC 50 of 27.25 Ϯ 2.74 M (38%) and a lower affinity IC 50 of 1064.54 Ϯ 79.11 M (62%) (Fig. 8B, inset).
Mibefradil was marketed by Roche Applied Science as a drug for treatment of hypertension and angina (18) before it was withdrawn in 1998 for its potential side effects. It is a nonselective antagonist but typically has an ϳ10-fold greater selectivity for T-type channels over L-type Ca 2ϩ channels. LCa v 3 is in the range of sensitivity to mibefradil (680 Ϯ 0.03 nM) as mammalian T-type channels (Fig. 8D). Caution must be heeded when directly comparing results from different studies as mibe-fradil is highly sensitive to the charge carrier, charge carrier concentration, and holding potential (19).

DISCUSSION
Introduction to LCa v 3-Here we describe the first in vitro expression characteristics of a snail homolog of mammalian T-type channels and, also remarkable, is that LCa v 3 is only one of two full-length cDNA sequences determined for non-mammalian T-types to date. cDNAs are assembled from predicted exons from a number of invertebrate sequenced genomes (e.g. Drosophila melanogaster Ca-␣1T-RB, accession no. NM_132068), but low homology outside of the conserved transmembrane domains indicates that the predicted transcript assemblies are likely erroneous when analyzed with multiple sequence alignments of cDNAs derived from mRNA (LCa v 3 and C. elegans cca-1B, human Ca v 3.1 to Ca v 3.3). LCa v 3 codes for a 322-kDa protein of 2886 amino acids, which is the largest protein of any reported four-repeat ion channel expressed to date, including 1.25 times larger than mammalian T-type channels and 1.5 times larger than cca-1B from C. elegans, the only other reported invertebrate cDNA coding for a T-type channel. Whether T-type channels in other phylogenetic groups are this large or possibly even larger is not known.
Permeation and the DDEE Selectivity Filter-The transmembrane regions are not responsible for most of the extra mass of LCa v 3 and include the highly conserved, voltage-sensor domain (S1 to S4) the outer helix (S5), the P-loop, and inner helix (S6) in all four repeat domains (7). A unique DDEE selectivity filter (20) and a gating brake (21) are two trademarks of T-types that distinguish them from the Ca v 1 and Ca v 2 Ca 2ϩ channel families. Flexible side chains of each domain harboring key glutamate residues (EEEE) contribute to the selectivity filter by extending into the permeation pathway, where they are expected to bridge Ca 2ϩ ions as they pass through the pore of high voltage-activated Ca v 1 and Ca v 2 channels. 3 A highly conserved aspartate residue upstream in the selectivity filter and adjacent to the glutamate residue in Domain II may serve to attract incoming Ca 2ϩ ions to the ion-selective pore of all Ca 2ϩ channels, according to modeling studies by B. Zhorov. 3 T-type channels are reported to have a lower Ca 2ϩ selectivity over monovalent cations as the estimated reversal potential is less positive than high voltage-activated Ca v 1 and Ca v 2 channels (ϩ40 versus ϩ60 mV) (22). Shortened carbon side chains in Domains I and II of T-types (DDEE instead of EEEE) may bridge Ca 2ϩ ions less stringently, resulting in lower pore selectivity for Ca 2ϩ ions in favor of faster kinetics that is typical for T-type Ca 2ϩ channels. Interestingly, inactivation kinetic changes mirror changes in activation kinetics in T-types (12), and a modified EEDD locus alters gating properties (20). No obvious conclusion can be drawn from differences in permeability for Ba 2ϩ and Ca 2ϩ ions among T-type channels. Only LCa v 3 and mammalian Ca v 3.2 channels have larger macroscopic Ba 2ϩ than Ca 2ϩ currents. Interestingly, greater macroscopic currents in Ba 2ϩ over Ca 2ϩ are a consistent feature with snail Ca 2ϩ channels expressed in HEK-293T cells including LCa v 1 (9) and LCa v 2 (23).
Gating Brake-A gating brake shared among T-type channels is considered to prevent channel opening at hyperpolarized potentials, as nucleotide polymorphisms in patients with childhood absence epilepsy or strategically placed deletions in this region produce channels that open at even more negative potentials than typical T-type channels (24). The proximal I-II loop of LCa v 3 is predicted to contain the helix-loop-helix gating brake structure (21), and more distally the I-II loop has been ascribed to regulating the surface expression of T-types (25). It may be more than coincidence that the gating brake is in the equivalent position where ␤ subunits associate with and alter the biophysical properties of high voltage-activated Ca v 1 and Ca v 2 channels as well as regulate/modulate their expression (e.g. protein folding, turnover, and membrane trafficking) (26). Indeed, invertebrate LCa v 3 does not require accessory ␤ or ␣2␦ subunits and robustly expresses in human HEK-293T cells at FIGURE 5. Invertebrate LCa v 3 has a large, persistent window current up to 1.8% of the total current near the resting membrane potential. A, sample current traces of maximal Ca v 3 currents (step to Ϫ35 mV) in response to a 1-s inactivating prepulse. B, a Boltzmann-fitted inactivation curve was generated by plotting the fraction of maximal current as a function of prepulse voltage. The fraction of maximal conductance at each voltage was plotted as an activation curve, curve-fitted with a Boltzmann function. The activation curve was derived from the current-voltage relationship minus the ohm-changes due to the driving force (illustrated in Fig 4D). C, calculation of the window currents were based on the product of the fraction of the whole cell conductance and fraction of available, non-inactivated channels at each voltage. Inset, a window current was measured at the end of a long, 1-s voltage-step. At 1 s, the majority of open channels will have been inactivated, leaving only open channels that persist under steady-state conditions, with a maximum at the resting membrane potential (Ϫ65 mV). FIGURE 6. Invertebrate LCa v 3 slowly deactivates similar to mammalian T-type channels. Sample tail currents and curve fitting of decay rate of tail currents (, ms) were generated by hyperpolarizing steps between Ϫ110 and Ϫ60 mV for 450 ms from a 7-ms depolarizing step to Ϫ35 mV.
an efficiency that rivals the mammalian T-type channels. Continued transfection in the presence of G418 antibiotic selection has generated a number of stable HEK-293T cell lines for LCa v 3. A high constitutive expression of LCa v 3 under the strong mammalian cytomegalovirus promoter argues in favor of greater transcriptional controls for T-type channel expression in native cells compared with perhaps more post-translational checkpoints regulating the expression of Ca v 1 and Ca v 2 channels that are known to form complex, multimeric assemblies along the secretory pathway.
Overall Shared Features of T-type Channels-Scoring of the overall amino acid conservation between invertebrate and mammalian genes can lead to overestimates of the degree of structural divergence, as a sequence not under selection will drift substantially over the hundreds of millions of years separating their evolution. Comparing the in vitro expression characteristics between LCa v 3 and mammalian T-types suggests a structural equivalency in core regions despite the overall sequence divergence of different channels, revealing a set of quintessential properties shared by all T-types. Voltage properties are tightly regulated with fast and transient kinetics, slow deactivation, window currents produced by overlapping activation, and availability curves and channel activity limited to a narrow window of subthreshold voltages where channels are available and conducting. Also, similar drug sensitivities of LCa v 3 for Ni 2ϩ ions and mibefradil suggest conserved residues in the outer pore and the aqueous permeation pathway between the selectivity filter and the aqueous, pore-lined, inner S6 helices (inverted tepee-shape) as predicted from the three-dimensional structure of crystallized K channels (27). Probing the affinity of a number of different T-type channel drugs will assist in interpreting the structural variants in the snail channel pore versus the mammalian ones.
Primitive Features in Invertebrate Channels-Invertebrate Ca 2ϩ channels of the high voltage variety are also highly conserved in their biophysical properties. Rat Ca v 1.2 and snail LCa v 1 channels, for example, are so alike that there are no reliable biophysical features outside of drug sensitivity that separate the two channels transfected in HEK-293T cells. 5 Differences outside of biophysical features appear to reflect the primitiveness of the invertebrate homologue, reminiscent of a state preceding the evolution of specializations in electromechanical coupling, such as the tetrad organization in skeletal muscle where mammalian Ca v 1.1 channels are directly coupled to ryanodine receptors of the sarcoplasmic reticulum (29). Invertebrate muscles lack tetrads or an equivalent Ca v 1.1 channel that mediates muscle contraction (29). More indirect coupling, with Ca 2ϩ serving as a short range transmitter, is also a feature of invertebrate neurotransmission. Invertebrate Ca v 2 channels that are responsible for transmitter release lack a II-III loop structure containing the synaptic protein binding site of Ca v 2.1 and Ca v 2.2 channels (11) and also exhibit a synaptic organization lacking key structural proteins present in mammalian synapses (such as Bassoon and CAST) and a synaptic substructure, such as a Drosophila T-bar, which is unlike the mammalian presynaptic density (30).
T-type Channel Diversity-T-type channels are modulated through intracellular signaling cascades and are coupled to other ion channels (31), but there is little to indicate that T-types serve as instruments for electromechanical coupling in cell-type specific, multisubunit complexes in the manner of Ca v 1 and Ca v 2 channels (32). Structural diversity in the three mammalian T-type channels arose out of genomic duplication, perhaps creating some overlapping redundancy in function. Yet the presence of unique biophysical properties, tissue specificity, modulation, and putative protein-protein interactions sites suggests otherwise, indicating that the different genes may provide specialized functions in mammals. Examples that illustrate this functional divergence include the contribution to rebound burst firing in thalamocortical neurons by Ca v 3.1, the involvement of Ca v 3.2 in pain sensitivity, relaxation of coronary arteries and secretion of aldosterone, and the involvement of Ca v 3.3 in long-lasting bursts in the inferior olive and habenula served by its slower kinetics and a larger window current range compared with other T-type channels (7). Distinct, regional antibody staining within individual central neurons suggests that each gene may serve particular roles within somatic, dendritic, and perinuclear compartments (33,34). Whether the diversity of mechanisms in mammals is contained within a single invertebrate Ca v 3 gene and its alternative splicing has not been explored.
Wide Range of Functions Expected for Abundant T-type Channel Transcript-Here we show that LCa v 3 is the most abundant Ca 2ϩ channel transcript in the Lymnaea nervous system, and our previous analysis indicates that this reflects a transcript profile in an individual snail neuron (11). Quantitative RT-PCR of single identified respiratory VD4 neurons, measured in replicates of six neurons, indicated that LCa v 3 is manyfold more abundant than either LCa v 1 or LCa v 2 channel expression (11). Their abundance in invertebrates may reflect a wide range of functions associated with T-type channels such as (a) shaping nerve action potentials and pacemaking, (b) a nonelectrogenic role for T-types in providing Ca 2ϩ through window currents (2), and (c) roles in differentiating and proliferating cells (3) and (d) secretion (35). Some invertebrates also appear to have additional roles that are not served by mammalian T-type channels, such as excitation contraction coupling in jellyfish muscle cells (36) or facilitating the contraction of pharyngeal muscles in nematodes (37). Interestingly, T-type spikes can provide qualitatively different information than sodium spikes in the same invertebrate axons. Weak depolarizations initiate slow swimming via T-type spikes, whereas stronger pacemaking inputs initiate a fast escape swimming response mediated by overshooting sodium spikes in the same axons, presumably operating in the availability range outside of T-type channels (38).
Drug Binding; Nickel-LCa v 3 has equal (ϳ300 M IC 50 ) Ni 2ϩ sensitivity as Ca v 3.1 and Ca v 3.3 channels. We report that the Ni 2ϩ dose-response curve for LCa v 3 is biphasic, indicative of two components of drug block. A similar biphasic Ni 2ϩ block is apparent in the dose-response data for mammalian recombinant channels (Fig. 3D in Ref. 16 39). A biphasic response might result from two Ni 2ϩ binding sites. Jones and co-workers suggest that Ca v 3.1 indeed has two binding sites, one in the outer pore and another deeper site within the pore pathway that is strongly affected by the permeant ion (42).
Unusually sensitive Ni 2ϩ block (5-10 M IC 50 ) is a property of Ca v 3.2 channels and critically involves a His-191 residue (16) in what has been described as the S3b-S4 voltage sensor paddle for sodium channels based on the x-ray structure of potassium channels (43). More than His-191 may be critical in the S3b-S4 voltage sensor paddle as a similar high affinity cation block of Ca v 3.2 channels by extracellular Zn 2ϩ involves the His-191 residue and two residues directly upstream of His-191, in particular, Asp-189 and Gly-190 (44).
The S3b-S4 is considered to carry most of the gating charge and likely drives the conformational changes required for pore opening and closing. It seems probable that Ni 2ϩ associates with the S3b-S4 paddle motif of Ca v 3.2 in a manner similar to how tarantula and scorpion toxins immobilize the voltage sensor of sodium channels (17). Ca v 3.2 is inhibited by Ni 2ϩ independently of voltage and is similarly blocked with Ca 2ϩ or Ba 2ϩ as a charge carrier, which is consistent with an inhibition by a mechanism outside the permeation pathway (42). Interestingly, LCa v 3 has an eight-amino acid insert in this short S3b-S4 region with extra positive and negative charges compared with Ca v 3 channels. The effect of the insert on Ni 2ϩ block or voltagegating, if any, is not known.
Other regions may also contribute to Ni 2ϩ block. High affinity Zn 2ϩ block in Cav3.2 channels also involves a neutral Ala-140 in IS2 that is negatively charged (Asp-140) in corresponding position of less sensitive Cav3.1 and Cav3.3 channels (44). Future chimera work may be important to evaluate whether LCa v 3 with a positively charged His-140 at this position influences cation block.
Drug Binding; Mibefradil-A number of new and potent T-type channel blockers are being explored, and mibefradil serves as the first T-type channel blocker that was clinically available (18). Interestingly, mibefradil block of snail LCa v 3 channel is in the range of potency of mammalian T-types. With doses spanning the mid-range of the IC 50 (680 Ϯ 0.03 nM), we observed that the mibefradil block of LCa v 3 would not readily stabilize, with accumulation of a slow but progressive block during long periods (tens of min) of continuous perfusion. We assume that this reflects a use-dependence often ascribed to mibefradil block (45). A slow time course of mibefradil block may also be explained by a reported accumulation of a hydrolyzed metabolite and more membrane-impermeant form of mibefradil (dm-mibefradil) that has an affinity for calcium channels from the cytoplasm (46). Further probing of different structures with mammalian and snail homologs will provide an opportunity for describing the high affinity drug binding in T-type channels.
Summary and Future Prospects-Expression characteristics of an invertebrate T-type channel has combined features that are reminiscent of all mammalian Ca v 3.1, Ca v 3.2, and Ca v 3.3 channels (see Table 1). LCa v 3 is 25% larger than any voltagegated ion channel expressed to date and is the most abundantly expressed Ca 2ϩ channel transcript in the snail nervous system. Window currents in invertebrate and mammalian T-type channels suggest a likely non-electrogenic role for T-types in providing Ca 2ϩ for proliferating and differentiating cells and in the developing embryo. Alternative splicing of the single invertebrate gene may provide the structural diversity for shaping the window current and firing patterns catered for individual network requirements. We anticipate that the snail will provide unique perspectives for probing T-type channel physiology. Much can be learned from the simple molluscan preparation where only a single T-type channel gene is expressed in native cells and where there is relative ease in probing the physiolog-ical mechanisms in single identified cultured neurons and intact networks in the brain that underlay well described behaviors (47). An invertebrate channel also provides an opportunity to reflect on evolutionary mechanisms. Channel I as it was first described has turned out to be the most challenging Ca 2ϩ channel to analyze since it was identified by Hagiwara et al. (5) more than 35 years ago. There is some truth in the following summary statement by Gray and Macdonald (48), reflecting on the present status of the T-type channel field: "[The] Physiologic regulation of T-type channels is simultaneously well documented and very obscure."