T-type Channels Become Highly Permeable to Sodium Ions Using an Alternative Extracellular Turret Region (S5-P) Outside the Selectivity Filter*♦

Background: Ion selectivity of voltage-gated channels is governed by selectivity filters. Results: Alternative turret region in domain II promotes highly sodium-permeable T-type channels without major changes to gating and kinetic features. Conclusion: T-type channels can generate variable sodium or calcium permeability by gene splicing. Significance: Ion selectivity in T-type channels can be altered using extracellular domains outside the ion selectivity filter. T-type (Cav3) channels are categorized as calcium channels, but invertebrate ones can be highly sodium-selective channels. We illustrate that the snail LCav3 T-type channel becomes highly sodium-permeable through exon splicing of an extracellular turret and descending helix in domain II of the four-domain Cav3 channel. Highly sodium-permeable T-type channels are generated without altering the invariant ring of charged residues in the selectivity filter that governs calcium selectivity in calcium channels. The highly sodium-permeant T-type channel expresses in the brain and is the only splice isoform expressed in the snail heart. This unique splicing of turret residues offers T-type channels a capacity to serve as a pacemaking sodium current in the primitive heart and brain in lieu of Nav1-type sodium channels and to substitute for voltage-gated sodium channels lacking in many invertebrates. T-type channels would also contribute substantially to sodium leak conductances at rest in invertebrates because of their large window currents.

T-type (Ca v 3) channels are categorized as calcium channels, but invertebrate ones can be highly sodium-selective channels. We illustrate that the snail LCa v 3 T-type channel becomes highly sodium-permeable through exon splicing of an extracellular turret and descending helix in domain II of the four-domain Ca v 3 channel. Highly sodium-permeable T-type channels are generated without altering the invariant ring of charged residues in the selectivity filter that governs calcium selectivity in calcium channels. The highly sodium-permeant T-type channel expresses in the brain and is the only splice isoform expressed in the snail heart. This unique splicing of turret residues offers T-type channels a capacity to serve as a pacemaking sodium current in the primitive heart and brain in lieu of Na v 1-type sodium channels and to substitute for voltage-gated sodium channels lacking in many invertebrates. T-type channels would also contribute substantially to sodium leak conductances at rest in invertebrates because of their large window currents.
Ca v 3 T-type channels are members of the 4ϫ6TM (4 heterologous domains of 6 trans-membrane segments) family of voltage-gated ion channels, which includes calcium (Ca v ) and sodium (Na v ) channels and sodium leak conductance channel (NALCN) (1)(2)(3). The four heterologous domains of 4ϫ6TM channels are considered to have evolved from two rounds of duplication of ancestral channels with one domain, such as the voltage-gated potassium channels (4). Key elements of each domain are a voltage sensor region (segments 1-4) and a re-entrant pore loop that extends from segments 5 and 6 and contains the selectivity filter thought to govern ion selectivity. The potassium channel pore mimics the hydration shell oxygen atoms that surround potassium ions in solution using conserved selectivity filter residues consisting of backbone carbonyl oxygens (5). This configuration makes it energetically feasible for surrogate oxygen groups to displace hydrating water molecules as potassium ions permeate (5). P-(Pore-) loops are contributed by four asymmetrical domains of calcium and sodium channels to a wider and shorter selectivity filter, which allows sodium ions to pass in a semi-hydrated (6) or hydrated state (7). Side chains of glutamate residues (EEEE) in the selectivity filter project carboxyl oxygens into the pore to create a high affinity pore locus for calcium ions in Ca v 1 (L-type) and Ca v 2 (non-L-type) calcium channels (8), where Na ϩ ions are prohibited from passing in the presence of low concentrations of external Ca 2ϩ ions (ϳ1 M) (9). T-type channels ubiquitously have a similar negatively charged ring of residues as Ca v 1 and Ca v 2 channels in the selectivity filter, but they are notably different, with aspartate residues replacing two of the glutamate residues in the 3rd and 4th positions (EEDD).
Mammalian T-type channels (Ca v 3.1, Ca v 3.2, and Ca v 3.3) have an ϳ10-fold lower calcium selectivity over sodium than Ca v 1 and Ca v 2 channels, but they are still impermeant to sodium in the presence of low (10 M) calcium (10,11). Here, we show that the LCa v 3 T-type channel from the pond snail Lymnaea stagnalis with a standard EEDD selectivity filter becomes highly permeant to monovalent cations through splicing of a novel extracellular turret, even in the presence of physiological (millimolar) concentrations of Ca 2ϩ ions. These sodium-permeant T-type currents serve as a major pacemaker current of primitive hearts, and are likely proxy for sodium channels when they are absent in many invertebrates. Our work confirms the capacity of snail channel currents to switch sodium and calcium permeabilities by an extracellular determinant, first postulated by Kostyuk et al. (12) more than 30 years ago.

Cloning and Expression of Novel LCa v 3 Exon 12a Splice
Variant-The original cloning and expression work of LCa v 3 T-type calcium channel containing exons ϩ8b, 12b, and Ϫ25c was described in Senatore and Spafford (13), and the original gene sequence was deposited as GenBank TM accession number AF484084. Subsequently, we described LCa v 3 T-type calcium channel splicing of exon 8b deposited as GenBank TM accession number JQ313138 and exon 25c deposited as GenBank TM accession number JQ313139. LCa v 3 Ϫ8b and ϩ25C splice variants were described in Senatore and Spafford (14). Here, we describe novel exon 12a isoform (ϩ8b, 12a, Ϫ25C), created by PCR using the original full-length clone of LCa v 3 in pIRES2 vector (ϩ8b, 12b, Ϫ25c) as template. The novel splice variant (ϩ8b, 12a, Ϫ25C) has been deposited as GenBank TM accession number JX292155. The inclusion of exons ϩ8b (54%) and 12a (56%) and Ϫ25C (77%) is the most common configuration of these three exons in the adult brain, whereas the adult heart does not commonly have a configuration of ϩ8b (16%) or Ϫ25C (21%) (14). The adult heart configuration of exons most commonly found is Ϫ8b (84%), 12a (99%), and ϩ25C (79%) (14). Novel exon 12a was confirmed in multiple PCR amplifications from snail brain cDNA (Fig. 1). All expressed plasmids were confirmed by sequencing (TCAG DNA Sequencing Facility, Sick Kids Hospital, Toronto, Canada). Human Ca v 3.1 expressible clone was a gift from Gerald Zamponi, University of Calgary, and cloned by Aaron Beedle.
Culturing of Snail Heart Cells-Heart ventricle cells were cultured from anesthetized adult snails (25-35 mm shell length) using a modified protocol (15). Cut up ventricles were placed in 0.5 mM Ca 2ϩ Leibowitz medium containing gentamycin (400 mg/ml) and 30 mM glucose, trypsinized (0.25% (w/v) trypsin, Sigma T9201) for 10 min, and then treated with 0.1% collagenase (Sigma type II) in 0.5 mM Ca 2ϩ Leibowitz medium for 30 min. The digested hearts were then washed three times with 3.5 mM Ca 2ϩ Leibowitz medium and then plated on acidetched circular glass coverslips (16) and left for 24 h to adhere at room temperature in 3.5 mM Ca 2ϩ Leibowitz medium containing 400 mg/ml gentamycin, 30 mM glucose, and 2% fetal bovine serum, before patch clamp recording 24 h later.
Whole Cell Patch Clamp Recording-We have detailed our optimized technique for the culture and maintenance of mammalian HEK-293T cells, calcium phosphate transfection and expression of ion channels, and their recording using whole cell patch clamp in an on-line video journal (17). There are five sets of solu-tions for recording (Tables 1 and 2). For comparing ion permeabilities or drug block, a Valvelink8.2 gravity flow Teflon perfusion system (AutoMate Scientific, Berkeley, CA) was used to toggle between differing external solutions during electrophysiological recording. Mibefradil was diluted in 5 mM water. A stock solution of isradipine was made in DMSO (dimethyl sulfoxide). Final concentration of isradipine was in 0.01% DMSO.
Whole cell patch clamp recordings were carried out with an AxoPatch 200B amplifier, combined with a Digidata 1440A data acquisition system and pCLAMP 10 software. Patch pipettes for recording had pipette resistances of 2-5 M⍀ (HEK-293T cells) or 5-10 M⍀ (heart cells) and with typical access resistance maintained after breakthrough between 4 and 6 M⍀ (HEK-293T cells) FIGURE 1. Expression of exons 12a and 12b in L. stagnalis cDNA. Top, PCR amplification of snail embryonic cDNA with nested primers flanking the mutually exclusive exons 12a and 12b produced two DNA fragments that could be distinguished in size on an agarose gel. Bottom, plasmid DNA inserts from A were excised from the pGEM-T Easy vector with EcoRI, and electrophoresed inserts were restricted to two sizes (451 bp for exon 12a and 484 bp for exon 12b, plus additional sizes for polylinker sequence from the pGEM vector). Sequencing of multiple inserts of each size confirmed that the smaller plasmid insert encoded exon 12a and the larger clone encoded exon 12b.  TEA-Cl HEPES  1  100  0  0  0  10  10  10  2  0  100  0  0  10  10  10  3  0  0  100  0  10  10  10  4 0 0 0 100 10 10 10 pH 7.2 with XOH where X= Cs, Na, K, or Li. # pH 8.0 with XOH or 10 and 14 M⍀ (heart cells). Only recordings with minimal leak (Ͻ10% of peak) and small current sizes (Ͻ500 pA) in HEK-293T cells were used due to loss of voltage clamp above 500 pA. Series resistance was compensated to 70% (prediction and correction, 10-s time lag). Off-line leak subtraction was carried out using the Clampfit 10.1 software (Molecular Devices). Protocols for measuring the voltage sensitivity and kinetics and curve fitting data are described in Refs. 13, 14. The relative permeability of P Ca /P x , where x is monovalent ion (Li ϩ , Na ϩ , K ϩ , and Cs ϩ ), was calculated by the following bi-ionic Equation 1 (18), Measurement of mRNA Expression of Exon 12a and Exon 12b Splice Variants Using qPCR-qPCR 4 was carried out as described previously (14,19,20). RNA was extracted from entire juvenile and adult snails, where sexually immature juveniles have shell lengths of 1.0 to 1.5 cm, and reproductioncapable adults have shell lengths of 2.0 to 2.5 cm (21). qPCR primer sets (Table 3) were designed to selectively amplify universal and specific exon 12a and exon 12b splice variants of Lymnaea Ca v 3. The specificity of PCR primer pairs was initially assessed by comparing the size of PCR products amplified from a pooled cDNA library with those amplified from cloned cDNAs. PCR primer efficiency for each primer set was then determined by generating relative standard curves using the 1:5 serial dilutions of pooled cDNA (1:5, 1:25, 1:125, and 1:625) as template for real time RT-PCR amplification. For each dilution, triplicate reactions were carried out in rigid 96-well PCR plates (Bio-Rad), with each well containing 0.5 l of serially diluted cDNA, 5 l of SsoFast TM EvaGreen Supermix (Bio-Rad), 0.5 l of each 10 M primer from a set, and 3 l of water. PCR amplification, fluorescence reading, and melt curve analyses were done using a Bio-Rad C1000 TM Thermal Cycler equipped with a CFX96TM Real-Time System and run by CFX Manager software (Bio-Rad). All cycle threshold values used for analysis were determined relative to the average cycle threshold value of the control gene, HPRT1 (hypoxanthine phosphoribosyltransferase 1).

Novel Spliced Turret Alters Monovalent Ion Permeability of
Snail T-type Channels-T-type channels diversified into three vertebrate genes but are rooted as a single T-type channel gene in extant relatives of animals spanning tissue and nervous system evolution (Trichoplax and cnidarians) and identified as the LCa v 3 gene in the pond snail, L. stagnalis (13,14). Expressible snail LCa v 3 channel resembles mammalian T-type channels in their rapid channel kinetics, and low threshold of activation for generating rhythmic pacemaker currents to threshold for eliciting action potential spikes (13). Snail channels also resemble particular mammalian T-type channels in their conserved and developmentally regulated exon splicing, such as with optional exons 8b and 25c, which confers membrane expression changes and changes to kinetic properties, respectively (14). We have identified unique exon splicing for exon 12 in the invertebrate T-type channel that spans from the S5 membrane helix of domain II, the extracellular loop, dubbed the "turret," to the

Sodium-permeable T-type Channels
descending pore helix that projects into the pore toward the selectivity filter ( Fig. 2A). Remarkably, the 3Ј-splice sites for exons 12a and 12b occur five amino acids upstream but do not include the invariant selectivity filter ( Fig. 2A), whose residues define the channel's characteristic ion selectivity (22). Exon 12a is shorter (39 aa long) with a characteristic cysteine structure, C xx C . . . C, whereas exon 12b has a penta-cysteine structure C . . . C x C . . . C x C and is longer (50 aa long) ( Fig. 2A). Snail LCa v 3-12a and LCa v 3-12b were recorded with 2 mM Ca 2ϩ as the external charge carrier in transfected HEK-293T cells, using whole cell patch clamp as described previously (13,14), with the cell culturing methods we describe in an on-line video publication (JoVE) (17). A notable difference for LCa v 3-12a compared with LCa v 3-12b in the ensemble of voltage-gated calcium currents was a large outward current with voltage steps above 0 mV (Fig. 2, B and D). We suspected that the large outward current with LCa v 3-12a was carried by the outflow of 110 mM Cs ϩ from the intracellular patch pipette, a constituent normally present in the patch pipette to block K ϩ currents. Replacement of 110 mM internal Cs ϩ with equimolar impermeant monovalent cation N-methyl-D-glucamine (NMDG ϩ ) eliminated most of the outward current for LCa v 3 isoforms, leaving a residual current, possibly carried by NMDG ϩ , but also the 0.6 mM Li ϩ in the compound Li-GTP and other monovalent contaminants contained in the intracellular solution (Fig. 2, C and D). Calcium channels have a high affinity for Ca 2ϩ ions, and monovalent ions are normally excluded in the presence of physiological levels of extracellular calcium, [Ca 2ϩ ] ex (10). Thus, it was surprising to observe that LCa v 3-12a generated Ͼ15-fold greater peak inward currents in the presence of 2 mM Ca 2ϩ with physiological levels of [Na ϩ ] ex at 135 mM, whereas LCa v 3-12b was significantly less permeable with only an ϳ2.5-fold increase (Fig. 2, E and F). With such a dramatic change in the pore's ion permeability for LCa v 3-12a, it is surprising that both snail LCa v 3-12a and -12b isoforms have completely overlapping biophysical properties, such as voltage sensitivities and kinetics of activation and inactivation (Fig. 3, A and B), as well as the size of the currents in divalent cations (Ba 2ϩ currents being ϳ1.3-fold larger than Ca 2ϩ currents) (Fig. 3C) and similar sensitivity to block by Ni 2ϩ ions (Fig. 3D). Outside of permeability changes, there is a notably faster recovery rate from inactivation ( Fig. 3E) and slower deactivation rate for the LCa v 3-12a isoform (Fig.  3F). Yet, taking it all together (see Table 4), it is surprising how an extracellular turret, constituting ϳ1% of the channel's length, has remarkably little effect on the T-type channel while dramatically altering monovalent cation permeability.
Snail T-type Channel with Exon 12a Is More Permeant to Sodium Ions than Calcium Ions-To quantify ion permeability differences, we measured peak sizes of T-type channel currents generated in bi-ionic conditions with intracellular solutions containing 100 mM of differing monovalent ions (Li ϩ , Na ϩ , K ϩ , or Cs ϩ ) in the presence of external calcium ions (4 mM [Ca 2ϩ ] ex ). We compared the more sodium-permeant snail LCa v 3-12a and snail LCa v 3-12b isoforms to the most sodiumimpermeant human T-type channel, Ca v 3.1. Sizes of outward currents were elicited from a Ϫ110-mV holding potential in voltage steps from Ϫ90 to ϩ80 mV in 5-mV increments (Fig.  4A). The relative permeability of monovalent ions followed the ion selectivity model by Eisenman et al. (23), where Li ϩ Ͼ Na ϩ Ͼ K ϩ Ͼ Cs ϩ , and is consistent with the decreasing permeability with increasing crystal radii of monovalent ions from Li ϩ Ͻ Na ϩ Ͻ K ϩ ϽCs ϩ (A o ) ϭ 0.60, 0.95, 1.33, and 1.69, respectively (Fig. 4, A and B). The contribution of each monovalent ion to a change in reversal potential reflects its relative permeability to the reference external calcium ion. Calculating the relative Li ϩ , Na ϩ , K ϩ , and Cs ϩ permeability in the bi-ionic Nernst equation reveals P Ca /P x with approximate averages of 23, 32, 49, and 118 for LCa v 3-12a channels, 26, 41, 78, and 114 for LCa v 3-12b channels, and 45, 90, 140, and 155 for human Ca v 3.1 channels (Fig. 5A). LCa v 3 channels with the exon 12a turret are significantly more permeable to monovalent ions than LCa v 3 channels bearing the exon 12b turret. Both snail channel splice variants are much more sodium-permeant than human Ca v 3.1, which is considered the most sodium-impermeant of the three mammalian T-type channels, Ca v 3.1, Ca v 3.2, and Ca v 3.3 (10,11). L-type channels are much more selective for calcium ions, and their monovalent ion permeability is barely detectable. L-type calcium channels are 10 -60-fold more selective for calcium ions than T-type channels with relative Li ϩ , Na ϩ , K ϩ , and Cs ϩ permeability ratios reported as P Ca /P x of 424, 1170, 3000, and 4200 (9,18). Another way to gauge the relative permeabilities is in the linear slope conductances between voltage steps from ϩ70 to ϩ80 mV, where the outward monovalent ion currents possess steeper conductance slopes in current voltage relations, reflecting the much greater observed outward monovalent current through snail LCa v 3-12a compared with snail LCa v 3-12b channels or human Ca v 3.1 (Fig. 5B). Monovalent ions such as sodium are competing with calcium ions for pore-binding sites (18), so we measured the changing peak amplitudes of the inward-permeating sodium current in 60 mM Na ϩ ex with 10-fold increment doses of extracellular calcium, in a range from 10 Ϫ9 to 10 Ϫ2 M. As calcium levels rose, there was an increased blocking of the Na ϩ current by calcium ions (Fig. 5C), reflecting a decrement of the sodium current with increasing external calcium. A striking difference for channels with the exon 12a turret is a weak calcium blocking effect (44.5 Ϯ 4.0%) compared with almost a complete block for channels with exon 12b turret (81.2 Ϯ 1.6%) or human Ca v 3.1 channels (96.1 Ϯ 1.6%) at 10 M [Ca 2ϩ ] ex (Fig. 5D). Above 10 M [Ca 2ϩ ] ex , Ca 2ϩ out-competes sodium in the pore and the observed in current rises through Ca v 3.1 channels as calcium reaches physiological (millimolar) levels ( Fig. 5E). Current sizes through Ca v 3.1 rise 9.1-fold from 10 M to 10 mM external calcium, in the presence of a constant 60 mM [Na ϩ ] ex reflecting the much greater permeability of the human Ca v 3.1 channel for calcium ions. The lack of calcium permeability of snail LCa v 3 channels is reflected in the 6.5-and 2.0-fold monotonic decline (LCa v 3-12a and LCa v 3-12b, respectively) in current size when calcium levels are rising between 10 M and 10 mM external calcium (Fig. 5, E and F). The switch from a calcium block of Na ϩ current to permeation and rise of calcium currents generates a characteristic U-shaped dose-response curve over the range of calcium concentrations for mammalian T-type channels ( Fig. 5C) (10). The consequence of the extracellular turret from exon 12a is to alter T-type channel ion selectivity, shifting a preference to passage of sodium ions over calcium ions espe-  Table 5.
cially at physiological concentrations and reducing the effectiveness of calcium ions to prevent sodium ions from permeating the channel pore. A summary of the monovalent and calcium permeability results are shown in Table 5.
Sodium-permeant T-type Channel Is the Only Major Sodium Current in the Snail Heart-We evaluated mRNA density of the differing extracellular turrets (exon 12a and 12b) in snail tissues by qPCR to determine where we could expect to find sodiumpermeable T-type channels. The highest density of T-type channels was detected in the snail heart, where the expression pattern of LCa v 3 is exclusively the sodium-permeant LCa v 3-12a isoform and where the exon 12b isoform is not detectable (Fig. 6A). Whole animal Ca v 3 transcript levels steeply decline during embryonic development, and LCa v 3-12a continues to fall from juvenile to adults (Fig. 6B), which likely is largely attributable to the sharp decline of LCa v 3 expression specifically in the heart (Fig. 6A). The sodium-impermeant isoform LCa v 3-12b is of equal abundance as LCa v 3-12a in the brain but is almost exclusive in secretory reproductive tissue (prostate and albumen gland), where LCa v 3 expression rises from juvenile to adult animals (Fig. 6A). We examined the relative density of mRNA for all relevant voltage-gated cation channels comparing the heart and brain (Fig. 6C). Snail hearts only express two cation channel genes, the L-type channel LCa v 1 (24) and the highly sodium permeant T-type channel LCa v 3-12a (Fig.  6C). Brain-enriched genes such as the classical sodium channel serving to generate the action potential upstroke (LNa v 1 (15)) is completely lacking in the snail heart. Also, the snail heart does not express the brain-enriched synaptic non-L-type calcium channel LCa v 2 (25). Note the line break in the y axis scale in Fig.  6C to accommodate the scale of the Na v 1 sodium channel expression in the snail brain (ϳ30-fold higher than control HPRT1 gene) that is dramatically above the range of expression of the other LCa v 1, LCa v 2, and LCa v 3 calcium channel genes (up to ϳ3-fold higher than control HPRT1 gene). The dramatic absence of expression of LCa v 3-12b and the highly brain-enriched LNa v 1 and LCa v 2 in the snail heart is evidence for the limited repertoire of four domain cation genes to LCa v 3-12a and LCa v 1 and a specialized signaling output in the snail heart that is highly different from the snail brain. We co-expressed and analyzed LCa v 1 (plus its required accessory ␤ and ␣ 2 ␦ subunits) and LCa v 3-12a channels in HEK-293T cells, which is a configuration expected to emulate the major cation channels in snail cardiomyocytes. Voltage ramps generated from Ϫ100 to

InacƟvaƟon Recovery
Values are expressed as mean +/-S.E.M. (numbers of cells used). TTP is the Ɵme to peak current. StaƟsƟcal comparisons were done using a one-way ANOVA combined with a Student-Newman-Keuls post hoc test with * P<0.05, ** P<0.01 and *** P< 0.001; n.s., not significant.   Fig. 6D). The high sodium permeation through LCa v 3-12a channels is revealed after replacing the 100 mM NMDG ϩ with equimolar Na ϩ (red curve, Fig. 6D). In the same recording, the HVA L-type current is unchanged when external NMDG ϩ is replaced with Na ϩ , reflecting the channel's barium-selective currents and a lack of sodium permeability for L-type channels (Fig. 6D). A voltage ramp recording of adult snail cardiomyocytes (Fig.  6F) generates a remarkably similar profile of the sodiumpermeant LVA T-type current and sodium-impermeant  Table 5.

InacƟvaƟon kineƟcs
HVA L-type current. The ϳ11-fold increase in size of the combined sodium and barium LVA current, upon addition of Na ϩ ions, is identical in cells transfected with LCa v 3-12a channels and the endogenous LVA T-type current in cardiomyocytes (Fig. 6H). On the contrary, the ϳ2-fold current increase in HEK cells transfected with LCa v 3-12b (Fig. 6, E and H) or human Ca v 3.1 (Fig. 6, G and H) is not reflected in the T-type currents of cardiomyocytes, corroborating the qPCR data that LCa v 3-12a is the exclusive T-type isoform in the snail heart. The similarity between cardiomyocytes and transfected LCa v 3-12a/LCa v 1 channels is also revealed in the makeup of current ensembles initiated from differing holding potentials (Fig. 7A). We used 2 mM Ba 2ϩ and 100 mM Na ϩ external solution to separate the kinetics of the slow barium L-type currents from the fast T-type currents mostly carried by sodium ions. A holding potential of Ϫ60 mV will inactivate T-type sodium currents, leaving a residual L-type current (gray color) and a mostly pure T-type sodium current (red color) from the difference current generated from Ϫ110 and Ϫ60 mV holding potentials (Fig. 7A). We examined drug responses of currents in cardiomyocytes elicited by voltage ramp (Fig. 7, B and C). Pharmacologically, we could separate the L-type from the T-type current using 1 M isradipine, but both the L-type and T-type currents were similarly blocked by nickel and mibefradil (Fig. 7, B and C). The sum of the electrophysiological data for cardiomyocytes (current kinetics, drug block, and sodium permeability) (Figs. 6 and 7) is consistent with the mRNA expression profiles in the heart,  Table 5. APRIL 25, 2014 • VOLUME 289 • NUMBER 17

JOURNAL OF BIOLOGICAL CHEMISTRY 11959
which indicate an exclusive expression of sodium-permeant LCa v 3-12a and sodium-impermeant LCa v 1 channels. Alternatively Spliced Turrets for Generating Sodium-permeable T-type Channels Are Widespread in Protostome Invertebrates-Analyses of Ca v 3 genomic sequences reveal that alternative splicing of exon 12 that we discovered in the snail L. stagnalis is widely present but limited to protostomes (i.e. nonechinoderm invertebrates) (Figs. 8 and 9 and Table 6). Exon 12a is characteristically short (38 -46 aa long, average ϭ 40.7 aa) with a nearly invariant tri-cysteine structure, CXXC . . . C (Fig.  8). Exon 12b has a penta-cysteine structure C . . . CXC . . . CXC (most protostomes) or CXXC . . . C . . . CXC (some nematodes) and is always longer (range, 48 -55 aa long; average ϭ 52.1 aa) than exon 12a (Fig. 9). The three vertebrate T-type channels, Ca v 3.1, Ca v 3.2, and Ca v 3.3, have an exon 12 with only one conserved cysteine (Fig. 8) compared with three and five cysteines in invertebrate exons 12a (Fig. 8) and 12b (Fig. 9), respectively. Paradoxically, the vertebrate exon 12 resembles the shorter exon 12a of invertebrates in size (38 -39 aa long, average ϭ 38.7 aa) which in snail LCa v 3 we have shown imparts the properties of extreme sodium ion permeability for exon 12a-containing T-type channels.

Unique Splicing of Exon 12 in Invertebrate T-type Channels-
We have functionally characterized the first nonvertebrate T-type calcium channel (13), which reveals a high conservation of quintessential features of vertebrate Ca v 3.1 and Ca v 3.2 T-type channels, including a similar low voltage range of activity with a peak current at Ϫ40 to Ϫ35 mV, with similar rapid activation and inactivation kinetics and slow deactivation kinetics. Also conserved is developmentally regulated splicing where T-type channels that lack exon 25c are highly downregulated from embryo to adult animals, leading to a battery of biophysical changes among LCa v 3, Ca v 3.1, and Ca v 3.2 (14). We also found that the presence of optional exon 8b in the I-II linker specifically down-regulates the level of protein expression of snail T-type channels in a manner consistent with Ca v 3.1 (14). Here, we describe unique exon splicing that generates highly sodium-permeant T-type channels utilizing an alternative exon 12a, which spans the extracellular turret of domain II upstream of the highly conserved EEDD selectivity filter of T-type channels.
We show that invertebrate LCa v 3 T-type channels will generate large outward currents in the presence of 100 mM intracellular monovalent ions accompanying the large driving force at high voltage steps much more so than human Ca v 3.1, while also generating an inward calcium current in the presence of 4 mM extracellular calcium ions at voltage steps below the reversal potential (Figs. 2 and 4). Permeability of monovalent ions decreases with increasing ion diameter size where Li ϩ Ͼ Na ϩ ϭ K ϩ Ͼ Cs ϩ Ͼ NMDG ϩ . All T-type channels pass sodium ions when external calcium concentration is at exquisitely low levels (Ͻ10 M external calcium ions). However, the sodium current through human Ca v 3.1 has less physiological relevance because it is almost completely blocked (96%) by low (10 M) external calcium. With rising external calcium concentrations to physiological levels, nearly all of the current through human Ca v 3.1 is carried by calcium (Fig. 5). The snail T-type channel isoforms, especially those harboring exon 12a, are not calcium-selective like human Ca v 3.1. Calcium ions are not effective at blocking the sodium current at 10 M external calcium, with snail T-type channels with exon 12a resisting the calcium block (44%), much more than isoforms with exon 12b (81%) or human Ca v 3.1 (96%) (Fig. 5). Increasing external Ca 2ϩ above 10 M to physiological (millimolar) levels does not generate the larger calcium currents for LCa v 3 as is reflected in the "U" shape calcium doseresponse curve of human Ca v 3.1 (10). Rather, the total current size dramatically falls monotonically (6.5-fold) for snail channels harboring exon 12a reflecting their reduced affinity for Ca 2ϩ in the pore that allows for greater permeability of Na ϩ ions (Fig. 5).
The extreme sodium permeability of LCa v 3-12a is evident in voltage ramp-generated currents with amplitudes that are ϳ11-fold greater when external Na ϩ replaces impermeant cation NMDG ϩ in the presence of barium. The sodium permeability is not evident in the co-expressed snail L-type (LCa v 1) channel in the same voltage ramp and a much reduced (ϳ2-fold instead of ϳ11-fold) sodium permeability for LCa v 3 channels harboring exon 12b (Fig. 6). Snail heart cells only express T-type channels with exon 12a and express the same sodiumpermeable T-type current as the in vitro expressed gene, resembling its fast kinetics, low voltage of activation, nickel (millimolar) and mibefradil (millimolar) sensitivity, and isradipine (micromolar) insensitivity (Figs. 6 and 7).
Unique Functions for Highly Sodium-permeant (Exon 12a) and Less Sodium-permeant (Exon 12b) T-type Channels-The distinct tissue expression patterns of LCa v 3 exons 12a (heart and muscle) and 12b (secretory glands) suggest specialized  3-12a 0.100 0.0057 8 0.079 0.0042 5 0.051 0.0037 12 0.027 0.0017 8 outward g X LCa v 3-2b 0.075 0.0046 3 0.060 0.0047 8 0.042 0.0032 5 0.014 0.  roles of different permeable T-type channels in different cell types. T-type channel transcript levels steeply decline during embryonic development through adulthood, reflected by a sharp decline in LCa v 3-12a expression in juvenile to adult heart (Fig. 6). A down-regulation of LCa v 3-12a may relate to allometric scaling during development, where more prominent LCa v 3 currents are found in smaller animals (non-adults) coinciding with faster heart rhythms (26). Expression of LCa v 3-12a isoform indicates a significantly slower deactivation rate and faster recovery rate from inactivation when compared with LCa v 3-12b (Fig. 3). This is consistent with LCa v 3-12a and adaptations in the snail heart, where a slower deactivation would serve to prolong the repolarization phase of the cardiac action potential, and a faster inactivation recovery prevents rundown between heart beats due to cumulative inactivation. Snails have only one (nonspecialized) sodium channel gene (Na v 1), compared with 10 in vertebrates, and this gene is only expressed in the snail brain and is not expressed in the snail heart (Fig. 6) (15). Thus, LCa v 3-12a is the only voltage-gated sodium channel generating pacemaker currents in ventricular cells of the primitive twochambered snail heart, which lacks the appearance of specialized vertebrate sodium channels, like Na v 1.5 and other neuronal Na v channels for fast conduction of heart rhythms. The more calcium-selective isoform LCa v 3-12b is of equal abundance to LCa v 3-12a in the brain, and it is the almost exclusive transcript associated with reproductive tissue of the hermaphroditic snail (prostate ϭ male organ and albumen gland ϭ female organ) (Fig. 6). The rise in transcripts of LCa v 3-12b from juvenile to adults (Fig. 6) coincides with sexual maturation where the albumen gland takes up a role secreting factors that facilitate egg mass formation (27). Our observations of a much faster deactivation rate and a slower recovery rate from inactivation of LCa v 3-12b channels promotes the typical pulsatile, burst firing of vertebrate T-type channels, where the rise in intracellular calcium may contribute to pacemaking and possibly also to excitation-secretion coupling (14).
Snail LCa v 3 (like its vertebrate counterparts) also has prominent "window" currents, where an estimated 1-2% of available channels are open at rest (13). LCa v 3-12a variants are highly abundant in the brain and likely contribute to reported persistent sodium currents (28,29) serving as a significant sodium leak conductance (30). Even though highly sodium-permeant T-type channels are likely widespread in invertebrates, it would be hard to discern them with voltage-gated sodium channel currents if you were not looking for them. Drugs are not selective in most invertebrates, like snails that are insensitive to classical sodium blockers such as tetradotoxin (15), and neither are snail T-type channels easily discernible by Ni 2 block (13). Some species, including nematodes like Caenorhabditis elegans, lack a sodium channel (Na v ) gene in their genomes (31), which means that their sodium-permeant T-type channels with exon 12a likely serve as a critical source for generating sodium-dependent spikes in these organisms. Most invertebrates do contain gene homologs of hyperpolarization-activated cyclic nucleotide-gated channels, which presumably contribute to sodium currents in invertebrate pacemaker rhythms (32). A hyperpolarization-activated cyclic nucleotide-gated channel homolog is also notably absent in C. elegans (31,32), which limits the possibilities of sodium current generating spikes in these animals from the conventional palette of ion channel genes. We suggest that T-type sodium channels have a role to play in lieu of sodium channels in C. elegans.
Evolutionary Adaptations of the Domain II Turret-A pattern of evolution of the extracellular turret in domain II of T-type channel is evident in their phylogeny (Fig. 10). The Ca v 3 T-type channel class first appeared in the most primitive multicellular animals, as low voltage-activated calcium-selective  counterpart to the Ca v 1 and Ca v 2 calcium channels. We have isolated and characterized the singleton T-type channel, TCa v 3, cloned from the most primitive, extant relative to have a T-type channel, the placozoan Trichoplax, which is multicellular but lacks a tissue level organization. TCa v 3 generates calcium currents consistent with other basal metazoans, such as the hydrozoan jellyfish where it serves as the only calcium current generated in muscle action potentials (33). All basal metazoans have short exon 12 turrets with a small number of cysteines (0 to 2). Insertion of an intron before exon 12 in primitive protostomes, the platyzoa (acoelomates), allowed a duplication of exon 12 to generate dual turrets, one highly sodium-permeant tri-cysteine turret (exon 12a) and a more calcium-selective penta-cysteine turret (exon 12b) in ecdysozan and lophotrochozoan protostomes. Associated with protostomes is the appearance of the coelom and internal organs such as the heart, which require fast sodium-dependent spikes to coordinate the activities of their body organs. Na v 1 channels in vertebrates expanded to 10 Na v 1 channel genes fulfilling a universal spike generator role inside and outside the brain. Invertebrates, however, have a singleton Na v 1 sodium channel gene, which is absent in many species (e.g. C. elegans) and is not abundant outside the nervous system (e.g. Lymnaea snail). Outfitting sodium-permeant pores in T-type channels is a likely adaptation in invertebrates in lieu of spikes generated by Na v 1 channels.

Importance of High Field Strength Site in Selectivity Filter
Governing Calcium and Sodium Permeation-Ion channels have evolved unique ion pores that are selective for the two major external cations, calcium ions and sodium ions. Flexible side chains of carboxylate oxygens from negatively charged residues are proposed to bind cations with high affinity, and permeation is facilitated by incoming ions attracted to the negatively charged extracellular surface of the pore (34). These FIGURE 8. Aligned amino acid sequences of exon 12A in metazoan Ca v 3 channel homologs. Exon 12a codes for novel extracellular turret residues from the middle of domain II, segment five through the descending pore helix to five amino acids upstream of selectivity filter glutamate (purple) residue, and invariant aspartate residue after it, conserved in all calcium channels (orange residue). Exon 12a is flanked by conserved charged residues (red), surrounding a set of variable amino acids, outside of very conserved cysteine (yellow) residues. Exon 12a is ϳ40 aa with a tri-cysteine motif (CXXC . . . C). Vertebrate T-type channels resemble the size of exon 12A in length (ϳ39 aa) but are missing the most downstream two cysteines of invertebrate exon 12a. Animals bearing T-type channels below the protostome invertebrates (placozoan and cnidarians) have zero or two cysteines in exon 12a and lack an exon 12b. Most protostome invertebrates possess an alternative exon 12b (see Fig. 9) turret in addition to exon 12a, and these are highlighted and outlined. . Aligned amino acid sequences of exon 12B found in protostome Ca v 3 channel homologs. Exons 12B code for novel extracellular turret residues from the middle of domain II, segment five through the descending pore helix to five amino acids upstream of selectivity filter glutamate (purple) residue, and invariant aspartate residue after it, conserved in all calcium channels (orange residue). 12b exons are flanked by conserved charged residues (red), surrounding a set of variable amino acids, outside of very conserved cysteine (yellow) residues. Exon 12b is ϳ52 aa with a penta-cysteine motif (C . . . CXC . . . CXC) or more rarely (CXXC . . . C . . CXC), in some nematodes. Most protostome invertebrates possess a shorter alternative exon 12a turret (see Fig. 8) in addition to longer exon 12b, and these are highlighted and outlined. charged residues form a "high field strength site" at the outer end of the selectivity filter in equivalent position of the selectivity filter TLESWSM in the re-entrant pore contributed by equal domains of the homomultimeric prokaryotic sodium channel, NavAb (6). Backbone carbonyl residues of the preceding two residues TLESWSM form the two innermost ion coordination sites for the sodium channel's selectivity filter (6). Replacement with three aspartates TLDDWSD in the selectivity filter transforms the sodium-selective pore into a calcium-selective one in prokaryotic channels (35,36). Significantly, sodium and calcium ions enter, reside, and exit through the selectivity filter in a semi-or fully hydrated form through a much broader pore than a potassium-selective pore, and where strategic positioning of a limited number of residues can tip the balance in favor of whether mostly sodium or calcium ions or a mixture of both ions permeate (35). The sodium and calcium pore is a departure from the long and narrow potassium selectivity filter (TVGYG), which is a Goldilocks' "just right" fit for the dehydrated potassium ion, which underlies the potassium channels' nearly exclusive selectivity for potassium ions (5).
Identity of the single amino acid base of the equivalent field strength glutamate residue in prokaryotic sodium channels serves as the basis for categorizing metazoan sodium and calcium channels. This position is contributed by four nonequivalent residues of domains I-IV, in four domain-containing channels that contain a lysine in the domains III or II as DEKA (or DKEA) for Nav1 sodium-selective channels and lack a lysine residue in the calcium-selective Ca v 1 (EEEE), Ca v 2 (EEEE), and primordial Na v 1 (DEEA) or invertebrate Na v 2 channels, DEEA (37). The importance of this residue position is evident in experiments where sodium channels take on the high calcium selectivity of calcium channels after replacement of the DEKA high field strength site position with EEEE and vice versa (8,38).
Yet it is clear that the high field strength site in the selectivity filter is not the only critical feature to govern permeation through T-type channels. All T-type channels have a unique high field strength site residue of EEDD, but when mutated to more resemble calcium-selective Ca v 1 and Ca v 2 channels (as EEED or EEDE), T-type channels become paradoxically more permeable and not less permeable to sodium ions (22). Despite the consistency in the EEDD high field strength site, T-type channels vary greatly in their sodium permeability that ranges from 20, 25, and 40% of the current for Ca v 3.1, Ca v 3.2, and Ca v 3.3 channels, respectively, for mammalian channels (1) and from 50 and Ͼ90% of the current through snail LCav3 channels with alternative novel extracellular S5-P turret regions in domain II. T-type channels also vary widely in their relative permeation of calcium ions as the charge carrier versus barium ions, which is expected to reflect a variable calcium affinity for the pores of T-type channels (10). This is in contrast to other calcium channels. Macroscopic barium currents are consistently ϳ2-fold the size of calcium currents, regardless of whether these are invertebrate Cav1 or Cav2 channels or vertebrate Cav1 or Cav2 homologs (1).
So, how do invertebrate T-type channels become so highly sodium-permeant using alternative splicing of extracellular regions outside the selectivity filter of the pore in domain II? Snail exons 12a and 12b differ in the extracellular turret, but there are also sequence differences that may be important in the descending helix just upstream of the selectivity filter in domain II. We suggest that changes in the descending helix and the extracellular turret could contribute to the altered sodium permeation through T-type channels.
Potential Roles for Changes in the Descending Helix-The distal end of the descending helix away from the selectivity filter has three variable amino acid residues as follows: AIV, ALI, and SLV in Cav3.1 and snail LCa v 3-12a and LCa v 3-12b channels, respectively. Negatively charged residues within the selectivity filter form transient binding partners outside the selectivity filter in structural intermediates of the calcium channel pore. FIGURE 10. Phylogeny and proposed evolution for exon 12, which governs monovalent ion permeability in Ca v 3 T-type channels. A, gene tree illustrates the conservation of exon 12a (smaller, blue exons) and exon 12b (larger, green exons) with vertical black lines indicating location of cysteine residues. Introns are indicated by triangles. B, proposed evolutionary progression; an intron separating exons 11 and 12; appearance of a tri-cysteine motif characteristic of exon 12a highly sodium-permeant isoforms; an alternative exon 12b with a penta-cysteine motif characteristic of less sodium-permeant channels in protostomes. Evolution of a uni-cysteine motif in exon 12 of Ca v 3 channels that supports a more calcium-selective channel in vertebrates is shown.
Changes in the descending helix may be sufficient to reorient a selectivity filter residue, altering the relative permeation of sodium and calcium ions.
Potential Similarities of the T-type Channel Turret with the Extracellular Cap Domain of K 2P Channels-There are possibilities of analogy between the extracellular turret of domain II in T-type channels and the crystal structures of turret regions from two-pore potassium leak channels (K 2P ) channels (40,41).
The first of two pores of K 2P channels has a uniquely extended turret of 56 amino acids, which normally ranges between 5 and 20 amino acids in other potassium channels. The turret forms a highly ordered extracellular cap, reminiscent of an A-frame, extending 35 Å above the lipid membrane with a cysteine residue at its apex, which cross-links to the apex of the A-frame of a second K 2P subunit, in the 2-fold symmetrical K 2P channel (40,41). The dual A-frames perpendicular to each other form a space-filled "carafe plug" that restricts ion and drug access to the channel pore from above (40,41). K ϩ ion passage is limited to two funnel-shaped side portals contained in the extracellular cap domain (40,41). Walls along the side portals are lined with negatively charged residues and variable sequences that could serve as a variable pre-filter for cations channeling to the selectivity filter below (40,41). The extracellular cap domain approaches close to the selectivity filter where the negative C-terminal ends of the helical dipole from an extracellular helix lie just above and to the side of the selectivity filter, before sharply pivoting away from the selectivity filter at a conserved glycine residue before meeting with the descending pore helix that leads to the selectivity filter (40,41).
The turret in domain II of T-type channels is the shortest of the four extracellular turrets preceding the pore helices that descend into the re-entrant pore and thus is likely restricted to the lower quadrant of the outer vestibule where it may likely encounter the selectivity filter (1). The constraints on size and the conserved framework of uni-, tri-, and penta-cysteine exon 12 in vertebrates and exons 12a and 12b of invertebrate T-type channels are consistent with these specialized turret variants changing the landscape in the external vestibule (analogously to K 2P channels) and forming regulatory structures that make specific contacts within the pore. Differing turrets are expected to bias the selectivity filter allowing the more favored passage of monovalent over divalent cations in a pore that is expected to be similar to the wide bacterial sodium channel (e.g. Na v Ab (6) and Na v Rh (42)), which can accommodate more than one semihydrated ion at its more constrictive locus (the selectivity filter residues).
Possible Similarity of the Extracellular Turret Domains in K 2P , K IR , and T-type Channels as Toxin Defense Shields-Eukaryotic inward-rectifying potassium channels (K IR ) also possess an extended turret region, albeit a minor external appendage compared with K 2P channels (43). Nonetheless, the turret of K IR channels is substantial enough to alter the external surface landscape and to contribute a unique resistance of these channels to classical invertebrate toxins (e.g. from snake, spider, and scorpion venom), which typically block voltage-gated potassium channels when applied from the outside of the channels (44 -46). Resistance to external toxins is a shared property with K 2P channels likely due to shielding with the elaborate extracel-lular cap domain of K 2P channels (47,48). Unique extracellular turret domains may also explain why there are not invertebrate pore-blocking toxins (from cone snail, spider, and snake venom) for Ca v 3 T-type channels compared with other (Ca v 1 and Ca v 2) calcium channels, where numbers of specific high affinity toxins have been discovered (49). The extensive extracellular turret regions in K 2P (40,41), eukaryotic K IR (43), and T-type channels are highly variable in sequence outside of key structural residues (e.g. cysteines, glycine, and prolines) or in the positioning of particular charged residues (e.g. arginine and lysine). One can imagine that some sequence variation in extracellular regions may be adaptive as counter-measures against the arsenal of specific channel toxins from the venom of invertebrate predators.
Other Examples of Alternative Splicing That Generates Altered Ion Selectivity of Ion Channels-Alternative splicing that generates permeation differences in ion channels is not common. We discovered alternative exons in domain II of NALCN channels that alter the high field strength residue to generate calcium-like (EEEE) or sodium-like (EKEE) pores (of flatworms, mollusks, annelids, echinoderms, and hemichordates) (19,50). Interestingly, some arthropods generate alternative pores in NALCN channels differently, using an alternative exon in domain III to generate calcium-like (EEEE) or sodium-like (EEKE) pores in their NALCN channels (19,50). Finally, a completely unique approach to altering ion selectivity is the use of alternative translational start sites to generate a truncated isoform in K 2P 2.1 channels that are sodium-permeable, to produce offsetting depolarizing currents in a channel type that typically only serves up hyperpolarizing potassium leak currents (39).
Conclusions-T-type channels have varying sodium permeability in the presence of an extracellular veil of novel turret residues in domain II, which encode merely ϳ1% of the T-type channel protein. Each splice isoform is exclusively expressed in particular tissues, where they serve highly different functions conducting variable sodium and calcium currents. It is surprising that only a few amino acids separate highly sodium-or more calcium-conducting channels, given that the ions have such different roles, where Na ϩ ions are relatively inert and are much more abundant serving mostly an electrogenic role, and Ca 2ϩ ions are maintained at very low levels in cells due to cytotoxicity, and serve as an exquisitely sensitive signaling molecule. Extracellular concentrations of sodium ions are on the scale of approximately fifty times that of Ca 2ϩ ions or K ϩ ions, so T-type channels that are permeant to monovalent ions are generating large currents and steeper membrane depolarizations than typical calcium-selective channels. They also generate larger currents than high voltage-activated calcium channels because of the greater driving force for T-type channels operating at their much lower voltage range.
It is remarkable that T-type channels evolved a mechanism outside the conventional selectivity filter to alter their ion selectivity. It is also intriguing to understand the physiological context (e.g. for brain, heart and glands) in which T-type channels retain similar biophysical properties but possess dramatically different ion selectivities for sodium and calcium ions.