Evolutionary insights into T-type Ca²⁺ channel structure, function, and ion selectivity from the Trichoplax adhaerens homologue

Voltage-gated calcium (Ca_v) channels play fundamental roles in the physiology of neurons and muscle, by coupling electrical signals carried largely by voltage-gated sodium (Na_v) and potassium (K_v) channels, with intracellular Ca²⁺-dependent processes (Clapham, 2007). Of the three classes of Ca_v channels, L-type/Ca_v1 channels are central for excitation-contraction coupling in muscle and excitation-transcription coupling in neurons and muscle, whereas N- and P-/Q-type (i.e., Ca_v2) channels are central for fast presynaptic excitation-secretion coupling (Catterall, 2011). T-type/Ca_v3 channels serve less obvious functions (Perez-Reyes, 2003; Senatore et al., 2012), but one clear contribution is their role in regulating cellular excitability, where their low voltages of activation and fast kinetics permit rapid depolarizing Ca²⁺ currents below the action potential threshold. T-type channels also play roles in driving low threshold exocytosis in both vertebrates and invertebrates, and in mammals have been shown to directly interact with presynaptic components of the vesicular SNARE complex (Weiss et al., 2012; Weiss and Zamponi, 2013). Notably, recent genomic studies indicate that T-type channels, and in fact the majority of genes with important roles in the nervous system, are present in primitive animals that lack nervous systems and single-celled organisms that predate animals (King et al., 2008; Srivastava et al., 2008, 2010; Steinmetz et al., 2012; Moran et al., 2015; Moroz and Kohn, 2015). We know little, however, about the function and properties of these extant gene homologues or about the functional or proteomic adaptations that were required to incorporate their primordial counterparts into nervous system function.

One very intriguing early diverging animal is Trichoplax adhaerens (phylum Placozoa), which has only six cell types and lacks synaptically connected neurons and muscle (Schierwater, 2005; Smith et al., 2014). Despite these absences, Trichoplax is able to coordinate motile behavior such as feeding (Smith et al., 2015), chemotaxis, and phototaxis (Heyland et al., 2014), indicative of trans-cellular signaling and communication independent of both chemical and electrical synapses. Given that Ca_v channels play crucial roles in both intra- and intercellular signaling, it is intriguing that the Trichoplax genome bears a full complement of Ca_v channel genes: Ca_v1, Ca_v2, and Ca_v3 (Srivastava et al., 2008).

Here, we sought to characterize the molecular properties of the most basal metazoan homologue of T-type channels from T. adhaerens. Co-immunolocalization of the channel, named TCa_v3, with neurosecretory cell marker complexin labeled gland cells, shown previously to resemble neurosecretory cells in their expression of SNARE proteins and the presence of membrane-apposed vesicles (Syed and Schierwater, 2002; Smith et al., 2014). We cloned and in vitro expressed TCa_v3, finding that despite its ancient divergence, it bears the hallmark structural and biophysical features of T-type channels, including a low voltage of activation, rapid and transient kinetics, and an apparent Ca²⁺ window current near resting membrane potential.

We also characterized the permeation properties of TCa_v3, finding that the channel conducts moderately mixed inward Ca²⁺-Na⁺ currents, with a majority of current carried by Ca²⁺, similar to mammalian homologues (Shcheglovitov et al., 2007). Paradoxically, measuring Ca²⁺ over Na⁺ selectivity using bi-ionic reversal potential analysis (i.e., where inward Ca²⁺ ions compete with outward Na⁺ for permeation), revealed poor Ca²⁺ versus Na⁺ selectivity compared with human Ca_v3.1, similar to the cloned T-type channel from invertebrate snail Lymnaea stagnalis (Senatore and Spafford, 2010; Senatore et al., 2014). We attribute the relatively low Na⁺ permeation through TCa_v3, in spite of its poor Ca²⁺ over Na⁺ selectivity, to retention of a potent Ca²⁺ block. Based on comparative data, we suggest that Ca²⁺ block is more crucial for determining the degree of Na⁺ that permeates alongside Ca²⁺ compared with pore selectivity and is a locus for evolutionary change in T-type channel cation permeability.

Two cDNA libraries were made from Trichoplax whole-animal total RNA, one with an anchored oligo-dT₁₈ primer, for PCR amplification and cloning of the C-terminal half of TCa_v3, and the other with a primer targeting a central region of the TCa_v3 coding sequence, for cloning the N terminus (Table 1). The TCa_v3 N- and C-terminal coding sequences were then independently amplified three times from the cDNA, via nested PCR using Pfu Turbo DNA polymerase (Agilent Technologies), with nested N- and C-terminal primer pairs containing NheI–XhoI and XhoI–XmaI sites, respectively. The nested NT primer (TCa_v3 NT 5′2) also contained a mammalian Kozak translation initiation site (Kozak, 1986; i.e., 5′-GCCACC-3′; Table 1) for effective expression of the TCa_v3 channel protein in mammalian cells. PCR-amplified DNA fragments were subcloned into pIRES2-IR–enhanced green fluorescent protein (EGFP), sequenced, and compared with each other plus the Trichoplax genome (JGI Genome Portal, Grell-BS-1999 v1.0, scaffold_2:6781672-6793175) to generate a consensus coding sequence. The full-length TCa_v3 clone was then prepared by inserting the XhoI–XmaI C-terminal subclone into the pIRES2 vector bearing the N-terminal TCa_v3 fragment, producing pTCa_v3-IR-EGFP. The full-length consensus coding sequence of TCa_v3 was submitted to GenBank (accession no. KJ466205).

The Trichoplax genome encodes single gene homologues for each of the three metazoan Ca_v channel types (Ca_v1, Ca_v2, and Ca_v3; NCBI accession nos. XM_002108894.1, XM_002109739.1, and KJ466205, respectively), as well as a single Ca_vβ accessory subunit gene (XM_002110305.1) and three Ca_vα₂δ Ca_v1/Ca_v2 accessory subunit genes (Ca_vα₂δa, Ca_vα₂δb, and Ca_vα₂δc; NCBI accession nos. XM_002112625.1, XM_002112621.1, and XM_002111347.1). Primers were designed to amplify ∼500-bp cDNA fragments of each of these genes by RT–PCR (Table 1), using a cDNA library prepared by RT (SuperScript III Reverse Transcription; Thermo Fisher Scientific) with an anchored oligo-dT₁₈ primer (Table 1) and whole-animal total RNA. PCR amplification was achieved in 25-µl reactions each containing 1.25 µl of 10 µM forward and reverse primers (Table 1); 0.125 µl of Taq DNA polymerase and 2.5 µl of corresponding 10× buffer (New England Biolabs, Inc.); 1 µl of 25 mM MgCl₂; 0.5 µl of 10 mM dNTP mix (New England Biolabs, Inc.); and 0.5 µl of cDNA template. Thermocycling conditions were 95°C for 2 min, 30 cycles of 94°C for 1 min, 59°C for 45 s, and 72°C for 1 min, and a final 10-min extension at 72°C.

Maximum likelihood (ML) phylogeny of various Ca_v channels was inferred from a MUSCLE-alignment of select channel protein sequences (Edgar, 2004), generated with the program MEGA7 (Kumar et al., 2016). Alignments were timed with trimAl (Capella-Gutiérrez et al., 2009), followed by some minor manual trimming to remove highly heterogeneous regions (raw sequences and the trimmed alignment are provided in FASTA format as Supplementary Files 1 and 2, respectively). ML model selection was achieved with MEGA7, revealing that the LG+G model was most suitable under both the corrected Akaike’s information criterion (AICc) and the Bayesian information criterion (BIC). The ML phylogenetic tree presented in Fig. 1 B was thus inferred from the trimmed alignment using the LG+G model, with 1,000 bootstrap replicates to generate node support values. Protein accession numbers used in the analysis are as follows: Amphimedon Ca_v1/Ca_v2, Aqu2.38198_001 from published transcriptome (Fernandez-Valverde et al., 2015); Salpingoeca Ca_v1/Ca_v2, XP_004989719.1; Salpingoeca Ca_v3, XP_004995501.1; Trichoplax Ca_v1 and Ca_v2, unpublished transcriptome; Trichoplax Ca_v3, KJ466205; Nematostella Ca_v1, XP_001639054.1; Nematostella Ca_v2a, Ca_v2b, and Ca_v2c, NVE4667, NVE18768, and NVE1263, respectively, from published transcriptome (Fredman et al., 2013); Nematostella Ca_v3a and Ca_v3b, NVE5017 and NVE7616, respectively, from published transcriptome (Fredman et al., 2013); Caenorhabditis elegans Ca_v1 (egl-19), NP_001023079.1; C. elegans Ca_v2 (unc-2), NP_001123176.1; C. elegans Ca_v3 (cca-1), CCD68017.1; Drosophila Ca_v1 (α1-D), AAF53504.1; Drosophila Ca_v2 (cacophony), AFH07350.1; Drosophila Ca_v3 (Ca-α1T), ABW09342.1; Lymnaea Ca_v1, AAO83839.1; Lymnaea Ca_v2, AAO83841.1; Lymnaea Ca_v3, AAO83843.2; human Ca_v1.1, NP_000060.2; human Ca_v1.2, Q13936.4; human Ca_v1.3, NP_001122312.1; human Ca_v1.4, NP_005174.2; human Ca_v2.1, O00555.2; human Ca_v2.2, NP_000709.1; human Ca_v2.3, NP_001192222.1; human Ca_v3.1, NP_061496.2; human Ca_v3.2, NP_066921.2; human Ca_v3.3, NP_066919.2; Mnemiopsis Ca_v2, fragmented transcriptome sequences manually pieced together from published transcriptome (Ryan et al., 2013); Hormiphora Ca_v2, sequence extracted from a de novo assembly of RNA-Seq data (SRR1992642; Francis et al., 2015); Chlamydomonas CAV2, XP_001701475.1; Schizosaccharomyces pombe CCH1, NP_593894.1; and Saccharomyces cerevisiae CCH1, NP_011733.3.

Trichoplax were frozen and freeze-substituted as described previously (Smith et al., 2014) with the following modifications. Coverslips (22 mm square, #1.5 thickness; ZEISS) were cleaned in nitric acid and treated with 3-aminopropyltriethoxysilane (#A3648; Sigma-Aldrich) to produce a positively charged surface. Trichoplax were transferred to a 500-µl drop of artificial seawater (ASW) placed in the center of the coverslips and left to adhere for 1–2 h. 300 µl of the ASW was removed and replaced with 500 µl of a 1:1 mixture of ASW and 1 M mannitol. After ∼5 min, the liquid was removed, and the coverslips were plunged into tetrahydrofuran at −80°C on dry ice and kept overnight. The coverslips were transferred to methanol with 1.6% paraformaldehyde on dry ice and then held at −20°C for 2–3 h and room temperature for 2 h. The specimens were rinsed in 100% ethanol (EtOH) and rehydrated gradually with 90%, 70%, and 50% EtOH (diluted with PBS) and PBS each for ∼10 min and blocking buffer (BB: 3% normal goat serum, 2% horse serum, 1% BSA in PBS) for 15 min. Then specimens were incubated in custom (Thermo Fisher Scientific) antiserum against the epitope ESRVNGNAKFTSDDQRLDR corresponding to the middle of the TCa_v3 I–II cytoplasmic linker (Fig. 1 A) or, to control for specificity, serum from the same rabbit before immunization both diluted 1:400 in BB overnight at 4°C. A custom (New England Peptide) chicken antibody against the epitope EATAPKKDSSKSNFSSR, found in the Trichoplax complexin protein, was added in some experiments to mark neurosecretory cells. After washing in PBS, the coverslips were incubated with Atto 488 goat anti–rabbit IgG (62197; Sigma-Aldrich) with/without Alexa Fluor 647 goat anti–chicken IgY (A-21449; Thermo Fisher Scientific) diluted 1:500 in BB for 2 h at room temperature. Nuclei were stained with Hoechst.

Images of immunostaining in Trichoplax were captured on an LSM 880 confocal microscope (ZEISS) with a 63× 1.4-N.A. PlanApo objective and 488-nm illumination for Atto 488 and 405-nm for Hoechst. Overview image stacks (17 images, 0.7-µm interval) were captured with a Quasar spectral detector with emission windows at 415–480 nm (blue) and 490–588 nm (green). Enhanced resolution image stacks (36 images, 0.185-µm interval) were collected with an Airyscan detector and 420–480- and 495–550-nm filters. Image stacks were displayed as maximum-intensity projections.

Trichoplax whole-animal protein lysates were prepared from ∼30 specimens and lysed directly in 200 µl of reducing sample buffer preheated to 95°C (50 mM dithiothreitol, 1% wt/vol SDS, 7.5% glycerol, 0.003% bromophenol blue, and 40 mM Tris pH 6.8). Protein lysates from ectopically expressed TCa_v3 channels in HEK-293T cells were prepared as follows. The entire coding sequence of TCa_v3 was excised from the pTCa_v3-IR-EGFP vector with restriction enzymes SacII and XmaI and cloned into pEGFP-C1 with the same sites (Takara Bio Inc.). The resulting plasmid pEGFP-TCa_v3, pTCa_v3-IR-EGFP, or the empty fusion vector pEGFP-C1 was cotransfected into HEK cells with rat Ca_vβ_1b and Ca_vα₂δ₁ subunits (as outlined in the culturing and transfection section of the Materials and methods below), and cells were incubated at 28°C for 4–5 d and then briefly washed with warm PBS (137 mM NaCl, 2.7 mM KCl, 10 mM Na₂HPO₄, and 1.8 mM KH₂PO₄, pH 7.4). Cells were then lysed with 600 µl of sample buffer (50 mM dithiothreitol, 1% wt/vol SDS, 7.5% glycerol, 0.003% bromophenol blue, and 40 mM Tris, pH 6.8). Equal volumes for each lysate were loaded on either freshly prepared 7.5% polyacrylamide gels or precast 4–20% polyacrylamide gradient gels (Invitrogen), and electrophoresis was performed in Invitrogen MES buffer using an XCell SureLock Mini-Cell Electrophoresis System (Invitrogen). For each experiment, paired gels were run: one was subjected to Coomassie staining to confirm equal protein content among samples, whereas the other was transferred to a nitrocellulose membrane using a Tris-glycine transfer buffer (25 mM Tris, 192 mM glycine, 20% vol/vol methanol, and 0.5% SDS, pH 8.3). Western blots performed using custom anti-TCa_v3 antibodies (rabbit polyclonal; Thermo Fisher Scientific) were done using (a) unpurified antibodies (terminal bleed serum, 1:1,000 dilution); (b) preimmune serum (1:1,000 dilution); or (c) affinity-purified antibodies (isolated by the manufacturer using a conjugated antigen peptide, 1:500 dilution). Western blots against EGFP epitopes (as an N-terminal fusion with TCa_v3 or alone) were performed using a rabbit polyclonal anti-GFP antibody (Sigma-Aldrich) at 1:5,000 dilution. Primary antibody incubations were performed overnight at 4°C, and secondary antibody incubations, washing, and detection were performed using standard chemiluminescent methods.

The detailed methods used for culture and CaPO₄ transfection of cloned Ca_v3 channels into HEK-293T cells, as well as techniques for whole-cell patch-clamp electrophysiology, have been previously documented in detail (Senatore et al., 2011, 2014; Senatore and Spafford, 2012). In brief, for electrophysiological experiments of in vitro–expressed TCa_v3, 6 µg of the pTCa_v3-IR-EGFP construct was transfected into HEK cells in 6-ml flasks, along with 3 µg of high voltage–activated calcium channel rat Ca_vβ_1b and Ca_vα₂δ₁ subunit cDNAs cloned into mammalian expression vector pMT2 (Tomlinson et al., 1993). These Ca_v1/Ca_v2 subunits have been shown to not interact with or alter the biophysical properties of heterologously expressed mammalian and invertebrate T-type channels (Dubel et al., 2004; Dawson et al., 2014), but nevertheless boost membrane expression by some unknown mechanism (Dubel et al., 2004). CaPO₄ transfections were done overnight at 37°C, after which cells were washed and transferred to 28°C for 2–3 d before recording. On the day of recording, cells were trypsinized (Sigma-Aldrich) and plated onto glass coverslips, which were then transferred into 2-ml culture dishes with appropriate extracellular recoding solutions.

For experiments involving the quantification of EGFP fluorescence in transfected HEK cells, transfections were performed in quadruplicate, with 6 µg pTCa_v3-IR-EGFP, pEGFP-TCa_v3, or pEGFP-C1 cotransfected with 3-µg combinations of rat Ca_vβ_1b and Ca_vα₂δ₁ subunits or the empty mammalian expression vector pCDNA 3.1. After incubation at 28°C for 3 d, the cells were imaged with transmitted and fluorescent light at 100 magnification, using a ZEISS AxioCam MRm Rev3 camera mounted on an AxioObserver A1 inverted microscope. All flasks were imaged with the same exposure settings, ensuring that the brightest cells did not saturate the pixels during acquisition with ZEN Lite software (ZEISS). Integrated density of the acquired fluorescence images was measured using the ImageJ software (Schneider et al., 2012), and values were normalized against the highest value for each replicate set, averaged, and plotted.

We used previously documented electrophysiological recording solutions to characterize the biophysical properties of TCa_v3 in 2 mM Ca²⁺ saline (Figs. 7 and 8; Senatore and Spafford, 2012), assess the Ni²⁺ block (Fig. 9; Senatore and Spafford, 2010), compare Ca²⁺ versus Ba²⁺ conductance (Fig. 10, C and D; Senatore and Spafford, 2012), and assess divalent versus monovalent cation selectivity and permeability features (Fig. 10, A, B, E, and F; Figs. 11 and 12; and Fig. S3; Senatore et al., 2014). Whole-cell patch voltage clamp recordings were performed using an Axopatch 200B amplifier and a Digidata 1440A digitizer controlled with pCLAMP 10 software (Molecular Devices). Pipettes were pulled using a Sutter P-97 micropipette puller and heat polished such that pipette resistance in the bath ranged from 2 to 5 MΩ, with access resistance after membrane breakthrough between 4 and 10 MΩ. Series resistance was not compensated for because we only kept data with minimal access resistance and tight capacitive transients upon voltage step. Only recordings in which leak current was <10% of the peak inward current were used, and offline leak subtraction was done using Clampfit software (Molecular Devices). Methods for Boltzmann transformation and curve fitting of electrophysiological data are described in previous publications (Senatore and Spafford, 2010, 2012). Relative permeabilities under bi-ionic conditions for TCa_v3 (i.e., P_Ca/P_X, where X = Li⁺, Na⁺, K⁺, or Cs⁺) were calculated using the bi-ionic Nernst equation (Hille, 2001) as described previously (Senatore et al., 2014). Statistical analyses comparing electrophysiology data for TCa_v3 with data from other in vitro–expressed channels were done using one-way analysis of variance (ANOVA); p-values are presented in Table 2, where we provide citations for data from other studies.

Fig. S1 shows the percentage minimum–maximum plot of the TCa_v3 channel protein coding sequence compared with the three human Ca_v3 channels (hCa_v3.1, hCa_v3.2, and hCa_v3.3). Fig. S2 is an alignment of the domain II P-loop of T-type channels corresponding to the exon 12 region. Fig. S3 shows superimposed current–voltage plots of various Ca_v3 channel homologues under different bi-ionic conditions. The Ca_v channel protein sequences used to generate the phylogenetic tree depicted in Fig. 1 B are provided in supplemental text file 1 (untrimmed, unaligned sequences) and supplemental text file 2 (trimmed, MUSCLE-aligned sequences), both in FASTA format.

Various Ca_v3 channel protein sequences were blasted against the T. adhaerens genome (Srivastava et al., 2008; JGI Genome Portal), identifying a predicted Trichoplax T-type/Ca_v3 channel homologue (T. adhaerens Grell-BS-1999 v1.0, scaffold_2:6781672-6793175). The predicted coding sequence served as a reference for RT-PCR amplification and sequencing of the Trichoplax Ca_v3 (TCa_v3) cDNA, amplified in two large fragments, with primers listed in Table 1. To build a consensus, every nucleotide along the 6,192-bp coding sequence of TCa_v3 was validated with a minimum of three independently amplified sequences. The resulting full-length open reading frame (submitted to GenBank with accession no. KJ466205) predicts a channel protein of 2,063 aa, a molecular mass of ∼238 kD, and a Kyte–Doolittle hydrophobicity profile with hydrophobic peaks corresponding to transmembrane helices (i.e., segments 1 to 6 or S1–S6) within each of IV repeat domains, conserved for all four-domain channels (Fig. 1 A). An inferred ML phylogeny of various Ca_v channel proteins places TCa_v3 basal to the two cnidarian T-type channels from Nematostella vectensis (Ca_v3a and Ca_v3b), as well as bilaterian protostome homologues C. elegans cca-1, Drosophila Ca-α1T, and Lymnaea LCa_v3, and chordate deuterostome homologues (human Ca_v3 isotypes Ca_v3.1-Ca_v3.3; Fig. 1 B). Recent phylogenomic studies have placed Trichoplax and its phylum (Placozoa) as a sister clade to the bilaterians and cnidarians, and sponges (phylum Porifera) and comb jellies (phylum Ctenophora) as the most early diverging animals (Srivastava et al., 2008, 2010; Ryan et al., 2013; Moroz et al., 2014; Pisani et al., 2015). Based on this phylogeny, Trichoplax is the earliest diverging animal to possess all three types of bilaterian/cnidarian Ca_v channels (i.e., Ca_v1, Ca_v2, and Ca_v3). Instead, the marine sponge/poriferan Amphimedon queenslandica and the two ctenophores Mnemiopsis leidyi and Hormiphora californiensis have only single Ca_v channel genes, forming either a clade with Ca_v2 types (i.e., ctenophores) or a sister clade with Ca_v1 and Ca_v2 types (i.e., Amphimedon, hence dubbed Ca_v1/Ca_v2-like; Moran and Zakon, 2014; Senatore et al., 2016). Interestingly, recent genome sequencing of choanoflagellate Salpingoeca rosetta revealed the presence of a T-type channel, indicating that Ca_v3 types likely predate Metazoa (Fig. 1 B; Fairclough et al., 2013; Moran and Zakon, 2014), and hence were lost in Porifera and Ctenophora. Thus, TCa_v3 is the most divergent homologue of vertebrate/human T-type channels identified to date in animals.

Of select Ca_v3 channels with validated mRNA sequences (i.e., TCa_v3, C. elegans Ca_v3 channel cca-1, human Ca_v3.1 to Ca_v3.3 isotypes, L. stagnalis Ca_v3, and Drosophila melanogaster Ca-α1T), the Trichoplax channel protein is among the smallest, with shorter cytoplasmic N- and C-terminal regions, as well as linkers between domains I–IV (Fig. 2 A). Instead, transmembrane regions (S1–S6 helices and corresponding linkers) are much more similar in length (Fig. 2 B), and indeed carry most of the protein sequence homology between different channels, whereas the cytoplasmic linkers and N/C termini exhibit much more divergence (Senatore and Spafford, 2010). The distant TCa_v3 bears what are perhaps the most distinguishing features of T-type channels: (a) a “selectivity filter” motif of EEDD, made up of negatively charged glutamate (E) and aspartate (D) residues that project into the pore to govern ion selectivity (Talavera and Nilius, 2006), distinct from the more calcium-selective EEEE selectivity filters of Ca_v1 and Ca_v2 channels (Figs. 1 A and 2 C); and (b) a predicted helix-loop-helix motif in the cytoplasmic linker between domains I and II, dubbed the “gating brake,” which serves to prevent channel opening at hyperpolarized membrane voltages (Perez-Reyes, 2010a) and where mutations in human Ca_v3.2 are associated with childhood absence epilepsy (Figs. 1 A and 2 D; Arias-Olguín et al., 2008).

Most of the ∼11,500 genes in the Trichoplax genome bear genomic architectures similar to orthologous genes in other animals (Srivastava et al., 2008). Accordingly, of the 28 exons/27 introns that make up the TCa_v3 channel gene, 26 splice junctions have counterparts in the Ca_v3.1 channel gene from mouse (MusCa_v3.1), whose mRNA transcript sequence is encoded by 38 exons/37 introns (Fig. 3). The Trichoplax Ca_v3 channel gene is ∼10-fold shorter than MusCa_v3.1, attributable to much smaller intron sizes here and across the entire Trichoplax genome (Srivastava et al., 2008). To summarize major structural differences between the Trichoplax and mouse Ca_v3 genes (Fig. 3): (a) TCa_v3 lacks an intron separating exons 2 and 3, in the coding region for the domain I S1-S2 linker (Fig. 1 A), but retains N-terminal introns separating exons 1 and 2, as well as exons 3–4, 4–5, and 5–6, which are conserved in all metazoan four-domain channels including Ca_v and Na_v channels (Spafford et al., 1999); (b) TCa_v3 also lacks an intron separating exons 11 and 12 encoding the domain II extracellular P-loop between S5 and S6, which is found in all Ca_v3 channels stemming from basal bilaterians (Senatore et al., 2014); (c) TCa_v3 lacks alternative donor splice sites at the 3′ ends of exons 8 and 25, which in snail LCa_v3 and MusCa_v3.1 create optional exons 8b and 25c that regulate channel surface expression and gating, respectively (Chemin et al., 2001; Emerick et al., 2006; Shcheglovitov et al., 2008; Senatore and Spafford, 2012); (d) TCa_v3 lacks exon 26 found in all mammalian Ca_v3.1 and Ca_v3.2 channel genes, as well as C. elegans cca-1, which produces similar but more slight gating effects compared with exon 25c (Chemin et al., 2001; Ohkubo et al., 2005; Steger et al., 2005; Zhong et al., 2006; Senatore and Spafford, 2012); and (e) TCa_v3 is missing four introns between exons 33 and 38 of MusCa_v3.1, corresponding to the channel C-terminal region. An overall reduction in genomic complexity for TCa_v3, especially in regions associated with alternative splicing, is consistent with reports that genes from basal metazoans generally undergo less alternative splicing (Pan et al., 2008; Wang et al., 2008; Gerstein et al., 2010; Graveley et al., 2011; Ramani et al., 2011). This was certainly evident during the sequencing and cloning of the TCa_v3 cDNA, in which we failed to identify a single alternatively spliced isoform, as well in an ongoing transcriptome analysis of Trichoplax whole-animal mRNAs (Senatore et al., 2016).

Adding to the analysis, Ca_v3 genes from premetazoan choanoflagellate S. rosetta and cnidarian N. vectensis reveal unique patterns in intron gain/loss across these different organismal lineages (Fig. 3). In keeping with the MusCa_v3.1 intro/exon numbering scheme, SalpCa_v3 exon 8 is fused with flanking exons 7 and 9, which are conserved as separate exons in mouse and Trichoplax, and bears an additional three internal exons. Notably, this region encodes the channel I–II linker protein sequence (Fig. 1 A), which tends to be highly divergent between different Ca_v3 channels. In this equivalent position, NemCa_v3a has an additional exon compared with mouse and Trichoplax, which, interestingly, overlaps with mouse optional exon 8b, which alters channel membrane expression. SalpCa_v3 notably lacks numerous introns between exons 12 and 33, a region encoding the C-terminal half of the channel protein (i.e., domains III and IV). Also evident is that MusCa_v3.1 exons 14–17 appear to have arisen via expansion from a single exon conserved in TCa_v3 and NemCa_v3a, in a region that corresponds the III–IV linker, which interestingly serves as a hotbed for modulation of mammalian T-type channels by kinases and G-proteins (Chemin et al., 2006; Perez-Reyes, 2010b; Senatore et al., 2012). Finally, NemCa_v3a and MusCa_v3.1 bear more exons/introns in the 3′ end of the gene, perhaps indicative of intron expansion in the cytoplasmic C terminus, a region with poor sequence homology where it is difficult to infer conserved intron/exon structure.

Immunostaining Trichoplax with both crude and affinity-purified custom antibodies against a I–II linker epitope of TCa_v3 (site depicted in Fig. 1 A) outlined cells around the rim of the animal and, less intensely marked, scattered cells further in the interior (Fig. 4 A). Enhanced imaging with the Airyscan detector (1.7-fold improvement in resolution and improved signal-to-noise relative to conventional confocal [Huff, 2015]) revealed that TCa_v3 staining was near the surfaces of hourglass-shaped cells and concentrated at the side facing the exterior of the animal (Fig. 4 A, inset). Staining was not evident when using preimmune serum (Fig. 4 B). The cells that labeled for TCa_v3 also were stained by an antibody against Trichoplax complexin (Fig. 4 C), a regulator of SNARE secretory proteins in neurons. The distribution of the cells as well as their shapes closely matched those of gland cells labeled with antibodies against the SNARE proteins syntaxin-1, synaptobrevin, and SNAP-25 (Smith et al., 2014). No signal was apparent in specimens incubated with preimmune serum and imaged with the same parameters.

We also confirmed expression of TCa_v3 at the mRNA level by RT-PCR, where gene-specific primers amplified an appropriate ∼500-bp fragment from cDNA reverse transcribed from whole-animal total RNA using an oligo-dT₁₈ primer (Table 1 and Fig. 5). Similarly, primers targeting the other two Trichoplax Ca_v channels, Ca_v1 and Ca_v2, as well as the Ca_v1/Ca_v2 accessory subunit Ca_vβ and three Ca_vα₂δ subunits (Ca_vα₂δ-a, Ca_vα₂δ-b, and Ca_vα₂δ-c; Table 1), all produced expected bands of ∼500 bp, with no bands evident when the RTase enzyme was omitted from the reaction (Fig. 5). We also note that mRNA expression of these genes, as well as other genes homologous to those involved in cellular excitability (e.g., Na_v2 and K_v channels, K⁺ leak channels), are expressed in the whole-animal transcriptome of Trichoplax (Senatore et al., 2016).

The TCa_v3 open reading frame (NCBI accession no. KJ466205) was cloned into bicistronic expression vector pIRES2-EGFP, which enables identification of positively transfected mammalian cells separately expressing TCa_v3 plus the fluorescent marker EGFP. Transfection of this construct (pTCa_v3-IR-EGFP) into HEK-293T cells produced barely detectable EGFP fluorescence (Fig. 6 A), suggesting that the TCa_v3 insert was somehow inhibiting expression of bicistronic EGFP because the empty pIRES2-EGFP vector normally produces robust EGFP fluorescence in these cells. Attempts to record voltage-activated Ca²⁺ currents from HEK cells transfected with pTCa_v3-IR-EGFP via whole-cell patch-clamp were unsuccessful, suggesting that the channel protein was also not expressed at high enough levels to accumulate at the cell membrane. A possible reason for the apparent poor expression of TCa_v3 is the prevalence of tandem rare codons in its cDNA sequence with respect to humans, which is expected to decrease efficiency of ectopic protein translation (Gustafsson et al., 2004; Kobayashi, 2015; Presnyak et al., 2015). Specifically, TCa_v3 exhibits a low codon adaptation index of 0.61 with respect to human preferred codons (Sharp and Li, 1987), and percentage minimum–maximum analysis reveals rare codon clustering along the entire length of the channel coding sequence (Fig. S1; Clarke and Clark, 2008).

Interestingly, previous research on cloned mammalian T-type channels revealed that coexpression with Ca_vβ and Ca_vα₂δ accessory subunits of high voltage–activated Ca_v1 and Ca_v2 channels increases Ca_v3 channel expression by an indirect mechanism (Dubel et al., 2004). Thus, we sought to increase TCa_v3 channel expression in HEK cells by cotransfecting with rat Ca_vβ_1b and Ca_vα₂δ₁ subunit cDNAs cloned into the mammalian expression vector pMT2 (Tomlinson et al., 1993). Strikingly, cotransfection of pTCa_v3-IR-EGFP with rat Ca_vβ_1b and Ca_vα₂δ₁ subunit vectors produced an ∼217-fold increase in bicistronic EGFP fluorescence compared with cotransfection with empty mammalian expression vector pCDNA-3.1 (Fig. 6 A), quantified as relative integrated density with ImageJ software (Fig. 6 D). Cotransfection with just Ca_vβ_1b or Ca_vα₂δ₁ separately also increased fluorescence, but more moderately, with respective increases of ∼165-fold and 148-fold (Fig. 6 D). In addition, cotransfection of pTCa_v3-IR-EGFP with the rat Ca_vβ_1b and Ca_vα₂δ₁ subunits allowed us to electrophysiologically record robust TCa_v3 Ca²⁺ currents in HEK cells.

Because bicistronic expression and fluorescence of EGFP from pTCa_v3-IR-EGFP only indirectly implies TCa_v3 channel protein expression, we repeated the cotransfection experiment using a construct in which the channel coding sequence was cloned in frame with that of EGFP at the channel’s N terminus, in the EGFP fusion vector pEGFP-C1. Cotransfection of the resulting pEGFP-TCa_v3 construct with rat Ca_vβ₁b and Ca_vα₂δ₁ subunit vectors increased EGFP fluorescence ∼75-fold, whereas Ca_vβ_1b and Ca_vα₂δ₁ alone increased fluorescence ∼47-fold and ∼38-fold, respectively (Fig. 6, B and D). Interestingly, the Ca_vβ_1b and Ca_vα₂δ₁ vectors also increased fluorescence of EGFP when transfected without TCa_v3, from the empty vector pEGFP-C1 (Fig. 6, C and D), and this corresponded with increased EGFP protein levels apparent on Western blots of corresponding HEK cell lysates probed with anti-EGFP antibodies (Fig. 6 F).

We were unable to detect the endogenous TCa_v3 protein in Western blots of Trichoplax whole-animal lysates or blots of protein lysates from TCa_v3-transfected HEK cells when using the custom anti-TCa_v3 antibodies. Instead, blots of HEK cells transfected with pEGFP-TCa_v3 plus the Ca_vβ_1b and Ca_vα₂δ₁ vectors produced appropriate bands of ∼270 kD when probed with anti-EGFP (Fig. 6 E), consistent with the predicted molecular weight of the TCa_v3 channel protein (238 kD) plus EGFP (32.7 kD). In accordance with our inability to record TCa_v3 Ca²⁺ currents when pTCa_v3-IR-EGFP was transfected without the Ca_vβ_1b and Ca_vα₂δ₁ subunit vectors, we were unable to detect the EGFP-TCa_v3 fusion protein without cotransfection of Ca_vβ_1b and Ca_vα₂δ₁ (Fig. 6 E, left lane). Thus, the rat Ca_vβ_1b and Ca_vα₂δ₁ subunit vectors appear to dramatically boost protein expression of TCa_v3, as either an EGFP fusion protein or a separate protein from the pTCa_v3-IR-EGFP construct. However, the effect of Ca_vβ_1b and Ca_vα₂δ₁ subunit vectors on ectopic protein expression appears to be at least in part nonspecific because they also boost expression of coexpressed EGFP in the absence of TCa_v3 (Fig. 6, D and F).

Finally, in lanes on Western blots in which the subunits were included and EGFP-TCa_v3 protein expression was evident, several additional bands could be observed with molecular weights of ∼100 kD and a triplet of intense bands near 33 kD (Fig. 6, E and F), suggesting that the channel either is being degraded in HEK cells or is incompletely translated. Application of proteasome inhibitor MG-132 to cells transfected with pEGFP-TCa_v3 plus pCDNA-3.1 for 12 h before fluorescence imaging caused only a 7.2-fold increase in EGFP integrated density (Fig. 6 G), still below the detection limit of Western blotting (not depicted), suggesting that proteasomal degradation only partly accounts for the poor expression of TCa_v3. Notable is that coexpression of EGFP-TCa_v3 with the Ca_vβ_1b and Ca_vα₂δ₁ subunits does not appear to increase channel protein expression by decreasing the amount of degradation, but rather, by increasing the total amount of protein, including the complete protein, plus all of the incomplete intermediates (Fig. 6, E and F).

Whole-cell voltage clamp recording of HEK cells cotransfected with pTCa_v3-IR-EGFP, rat Ca_vβ_1b, and rat Ca_vα₂δ₁ produced low voltage–activated calcium currents in 2 mM external Ca²⁺ in response to depolarizing voltage steps from −110 mV to between −90 and 40 mV (Fig. 7 A). A plot of peak currents versus step potential (i.e., current–voltage or IV plot) reveals a slightly hyperpolarized maximal peak inward current of −45 mV (Fig. 7 B), which is between 5 and 20 mV more negative than that of other Ca_v3 channels (Table 2). Boltzmann transformation of the IV plot, which removes the influence of driving force to estimate the voltage dependence for channel activation, indicates that TCa_v3 activation begins at very low depolarizing potentials compared with published data for other Ca_v3 channels derived using similar methods (i.e., compare half-maximal activation [V_1/2] of −59.32 ± 0.9 mV for TCa_v3 vs. −53.48 ± 0.34 for Lymnaea Ca_v3 [Senatore and Spafford, 2012]; −49.3 ± 0.7 for human Ca_v3.1 [Chemin et al., 2002]; −48.4 ± 1.2 for Ca_v3.2 [Chemin et al., 2002]; and −41.5 ± 1.1 for human Ca_v3.3 [Chemin et al., 2002]), reaching maximal activation near −40 mV (Fig. 7 C and Table 2). Conversely, steady-state inactivation of TCa_v3, determined by measuring peak residual current after exposure to prolonged steady-state voltages (Fig. 7 C, inset), is quite similar for TCa_v3 with respect to other channels, especially Ca_v3.1 and Ca_v3.2 (i.e., V_1/2 for inactivation is −74.15 ± 0.90 for TCa_v3 and −74.2 ± 1.1 for human Ca_v3.1 [Chemin et al., 2002] and −75.6 ± 0.7 for human Ca_v3.2 [Chemin et al., 2002]; Table 2). Altogether, the voltage properties of TCa_v3 indicate that it is likely more active at threshold voltages compared with other Ca_v3 channels, where the channel is subject to roughly the same amount of inactivation, but is more readily activated by depolarization.

An important and characteristic feature of T-type channels are their “window” currents, which occur at steady-state voltages near rest through a pool of constitutively open channels (Dreyfus et al., 2010), providing a constant influx of Ca²⁺ that depolarizes the cell membrane to alter cellular excitability (Cain and Snutch, 2010) and increases cytosolic Ca²⁺ to regulate cell growth and proliferation (Lory et al., 2006; Taylor et al., 2008; Senatore et al., 2012; Gackière et al., 2013). Such a window current is evident for TCa_v3, for which an overlap between channel activation and inactivation reveals a voltage range between −60 and −75 mV, where not all channels are inactivated and some degree of activation takes place (Fig. 7 C, red fill).

In vitro TCa_v3 currents exhibit slow onset (activation) and attenuation (inactivation) at slight depolarizing voltage steps, which accelerate with stronger depolarization “tightening” current waveforms, a hallmark of Ca_v3 channels that produces a crossing over of inactivation curves toward peak inward current (Fig. 7 A). Such changes in current waveforms can be quantified with time constants (τ) for mono-exponential curve fits over the rise (τ_activation) and decay (τ_inactivation) phases of channel currents, which for TCa_v3 results in a decrease in τ consistent with accelerating kinetics upon stronger depolarization (Fig. 7 D). The rate at which TCa_v3 activation accelerates through depolarization is lower compared with other in vitro–expressed channels, with only a 4.13-fold drop in τ_activation from −50 to −10 mV versus 4.41-fold for Lymnaea Ca_v3 (Senatore and Spafford, 2010), 5.50-fold for human Ca_v3.2 (Gomora et al., 2002), 7.29-fold for human Ca_v3.3 (Gomora et al., 2002), and 7.45-fold for human Ca_v3.1 (Fig. 7 E and Table 2; Gomora et al., 2002). Likewise, acceleration of inactivation kinetics is slower, with only a 1.29-fold decrease in τ_inactivation for TCa_v3 versus 1.76-fold for LCa_v3, 1.87-fold for hCa_v3.2, 1.58-fold for hCa_v3.3, and 3.88-fold for hCa_v3.1 (Fig. 7 E and Table 2). From this data, it is clear that the two invertebrate channels compared here, TCa_v3 and the T-type channel from mollusk L. stagnalis (LCa_v3), have kinetics with an overall reduced voltage dependency, especially for current activation. TCa_v3 current kinetics are marginally slower across all negative voltages than other in vitro–expressed channels, with the exception of the slow Ca_v3.3 channel (Table 2).

Similar to Lymnaea Ca_v3, and in contrast to mammalian Ca_v3 channels, TCa_v3 recovers slowly from inactivation (Fig. 8, A and B; and Table 2), indicating that prolonged hyperpolarization would be required to effectively recruit the channel from depolarized membrane voltages. TCa_v3 also has slow deactivation kinetics relative to mammalian channels at voltages near −70 mV (Fig. 8, C and D; and Table 2), which during action potential repolarization would result in increased inward Ca²⁺ influx. Overall, despite the noted differences, the voltage dependencies and kinetics of TCa_v3 currents are remarkably similar to those of Ca_v3 channels from animals that have neurons and muscle (Table 2), in particular where a low voltage of activation and rapid activation and inactivation kinetics (which allow T-type channels to contribute depolarizing currents near action potential threshold) are conserved.

Early electrophysiological experiments revealed that the divalent cation Ni²⁺ could potently block low voltage–activated Ca²⁺ currents in some vertebrate preparations (Perez-Reyes, 2003). This high-affinity block was subsequently attributed only to the Ca_v3.2 channel isotype, and specifically to a unique histidine residue in its domain I S3–S4 extracellular loop (His-191) that strongly binds Ni²⁺ to disrupt channel gating (Fig. 9 A; Kang et al., 2006, 2010). The recently cloned T-type channel from Drosophila also bears a histidine in this loop (albeit 2 aa positions upstream of Ca_v3.2 His-191) and, not surprisingly, is also highly sensitive to Ni²⁺ (Jeong et al., 2015). Instead, TCa_v3, mammalian Ca_v3.1 and Ca_v3.3 channels, and Lymnaea LCa_v3, all lack histidines in this region (Fig. 9 A). Accordingly, all of these channels have low and remarkably similar IC₅₀ values for Ni²⁺ block: 335.0 ± 6.5 µM for TCa_v3 (Fig. 9, B–D); 300.0 ± 29.2 for LCa_v3 (Senatore and Spafford, 2010); 304.8 ± 6.2 for human Ca_v3.1 (Kang et al., 2006); and 216 ± 9 for human Ca_v3.3 (Table 2; Kang et al., 2006). The similarity in IC₅₀ values for all of these channels suggests that low-affinity Ni²⁺ block occurs through a common mechanism, where perhaps, as has been suggested for Ca_v3.1, extracellular Ni²⁺ ions bind two distinct regions of the pore in a cooperative manner (Obejero-Paz et al., 2008). Such a model is perhaps applicable to TCa_v3, where the Hill coefficient for Ni²⁺ block is greater than 1 (i.e., 1.18 ± 0.03; Fig. 9 D), suggesting some degree of cooperative binding. Interestingly, washout of Ni²⁺ is particularly fast for TCa_v3 compared with other T-type channels (Kang et al., 2006; Senatore and Spafford, 2010), with a transient increase in peak current amplitude (I/I_max) after perfusion of extracellular Ni²⁺ is replaced with Ni²⁺-free saline (Fig. 9 C).

Recently, protostome invertebrates were found to uniquely possess alternative exons 12a and 12b, encoding alternate turret and descending helices of the domain II pore-loop (P-loop). In the freshwater mollusk L. stagnalis, these exons were found to produce channels with extremely bifurcated permeability features: whereas channels with exon 12b (i.e., LCa_v3-12b) conduct moderately mixed Ca²⁺-Na⁺ currents under physiological conditions, LCa_v3-12a is extremely Na⁺ permeant, such that less than 10% of inward current is carried by Ca²⁺ (Senatore et al., 2014). Alignment of the domain II turrets of Ca_v3 channels from basal metazoans Trichoplax (TCa_v3) and N. vectensis (Ca_v3a) with channels from protostomes (i.e., L. stagnalis, Drosophila, and C. elegans), deuterostomes (Ciona intestinalis, human), and a premetazoan species (choanoflagellate S. rosetta) reveals that the TCa_v3 domain II P-loop resembles those of exon 12a–bearing channels, being smaller and containing fewer cysteines than exon 12b (Fig. S2).

We sought to assess whether the exon 12a–like turret of TCa_v3 is associated with moderate Na⁺ permeability, similar to cnidarian (Lin and Spencer, 2001), mammalian (Shcheglovitov et al., 2007), and basal deuterostome (Hagiwara et al., 1975) channels, or instead associates with extreme Na⁺ permeability, similar to exon 12a variants of Lymnaea Ca_v3. For this, we assessed the degree of mixing of Ca²⁺ and Na⁺ currents through TCa_v3 ectopically expressed in HEK-293T cells by replacing a perfused extracellular recording solution containing 2 mM Ca²⁺ and 135 mM impermeant NMDG⁺, with one containing Na⁺ instead of NMDG⁺ (Fig. 10, A and B). Addition of Na⁺ resulted in a 42% increase in peak inward current elicited by stepping from −110 to −45 mV, which approximates the degree of Na⁺ expected to move through the channel alongside Ca²⁺ upon membrane depolarization (Shcheglovitov et al., 2007; Senatore et al., 2014). Comparing the current increase of TCa_v3 with previously published data of other cloned channels (Senatore et al., 2014; Stephens et al., 2015) reveals that the Trichoplax T-type is most similar to the least Ca²⁺-selective mammalian isotype Ca_v3.3 (i.e., 45%), whereas the exon 12b variant of LCa_v3 exhibits a larger increase of 153% in current amplitude and human Ca_v3.1 an increase of only 27%. In stark contrast, LCa_v3-12a undergoes a striking 1,440% increase in current amplitude upon perfusion of external Na⁺ (Fig. 10 B), reflecting its dramatic Na⁺ permeability. Thus, in the presence of Ca²⁺, TCa_v3 appears to conduct only moderately mixed Ca²⁺-Na⁺ currents, in a range similar to that of other in vitro–expressed T-type channels, with the exception of the highly Na⁺-permeant LCa_v3-12a.

T-type channels are highly permeable to divalent cation Ba²⁺, where for reasons that are unclear, the three rat Ca_v3 channel isotypes have dissimilarities with respect to macroscopic Ca²⁺ versus Ba²⁺currents in vitro (Talavera and Nilius, 2006): Ca_v3.1 conducts larger Ca²⁺ currents, Ca_v3.2 conducts larger Ba²⁺ currents, and Ca_v3.3 conducts equal Ca²⁺ and Ba²⁺ currents (McRory et al., 2001). Here, we found TCa_v3 to be similar to rat Ca_v3.1, where replacing 2 mM external Ba²⁺ with equimolar Ca²⁺ by perfusion produced an increase in current amplitude upon membrane depolarization from −110 mV (Fig. 10 C), with a 2.29-fold increase in maximal peak inward current visible on IV plots (Fig. 10 D). It appears as though the difference in macroscopic Ca²⁺ versus Ba²⁺ current amplitude for TCa_v3 is not attributable to differences in activation/inactivation kinetics in the presence of either ion, as was shown for rat Ca_v3.1 (Khan et al., 2008), where mono-exponential curves fitted over activation and inactivation phases of macroscopic currents for Ba²⁺ versus Ca²⁺ produce statistically indistinguishable τ time constants across all depolarizing voltages (Fig. 10 C). Interestingly, the Lymnaea T-type channel, which can alter its Ca²⁺ versus Na⁺ permeation properties via alternative splicing of exons 12a and 12b, nevertheless always conducts approximately twofold larger Ba²⁺ than Ca²⁺ currents in vitro, regardless of exon 12 splicing (Senatore and Spafford, 2010; Senatore et al., 2014).

Differences in Ca²⁺ versus Ba²⁺ permeation among the various in vitro–expressed T-type channels have no consequence for the pattern in fold increases in peak inward current upon addition of external Na⁺. Comparing the fold increase in peak macroscopic current through TCa_v3 elicited by ramping the voltage from −110 to 100 mV in the presence of 2 mM Ba²⁺ and either 100 mM Na⁺ or NMDG⁺ (Fig. 10 E) with previously published data from other in vitro–expressed channels (Senatore et al., 2014; Stephens et al., 2015) reveals a similar ranking in Na⁺ permeation as observed in the presence of 2 mM external Ca²⁺: LCa_v3-12a >> LCa_v3-12b > TCa_v3 ∼ hCa_v3.3 > hCa_v3.2 > hCa_v3.1 (i.e., compare Fig. 10, B and F).

At extremely low concentrations of external Ca²⁺, all Ca_v channels conduct prominent Na⁺ currents. Titrating increasing concentrations of extracellular Ca²⁺ leads to a block of Na⁺ current, with a sensitivity that reflects the affinity of Ca²⁺ to binding sites located in the pore. As [Ca²⁺]_out increases, the blocking effect reaches saturation levels, then Ca²⁺ itself becomes abundant enough to occupy multiple sites within the pore, leading to Ca²⁺ permeation (Tsien et al., 1987; Sather and McCleskey, 2003; Cheng et al., 2010; Buraei et al., 2014; Tang et al., 2014). This property, of a decrease in Na⁺ current amplitude and a subsequent rise in Ca²⁺ current as [Ca²⁺]_out increases (known as the Ca²⁺ block effect), is exemplified by peak currents through human Ca_v3.1 elicited by repeating voltage steps from −110 to −35 mV, whereas external Ca²⁺ is perfused from 1 nM to 10 mM in the presence of invariant 60 mM external Na⁺ (Fig. 11 A; Senatore et al., 2014). Notably, the TCa_v3 Ca²⁺-block data are similar to previously published data for human Ca_v3.1 (Senatore et al., 2014; Stephens et al., 2015), where at 10 µM [Ca²⁺]_out, 97.3% and 96.1% of peak Na⁺ current elicited by voltage steps to −35 mV were blocked, respectively (Fig. 11, A and B). Human Ca_v3.1 Na⁺ currents appear to be slightly more sensitive to Ca²⁺, where at 0.1 µM [Ca²⁺]_out, 34.1% of Na⁺ current was blocked, whereas TCa_v3 and the two exon 12 variants of LCa_v3 only exhibit between −5.6 and 4.6% block. Beyond 10 µM [Ca²⁺]_out, TCa_v3 and Ca_v3.1 show 13.8- and 9.4-fold increases in peak current caused by an emerging Ca²⁺ conductance (Fig. 11, A and C). In contrast, LCa_v3-12b shows a moderate reduction in Ca²⁺ block (82.6%), whereas the highly sodium-permeant LCa_v3-12a has a dramatically right-shifted dose–response for Ca²⁺ (Fig. 11 A), with only 44.5% of current blocked at 10 µM Ca²⁺ (Fig. 11 B; Senatore et al., 2014). Also noticeable is a continued decline in current amplitude beyond 10 µM [Ca²⁺]_out for the Lymnaea channel variants (Fig. 11 A), most marked for LCa_v3-12a, with a sevenfold continued decrease in current from 10 µM to 10 mM external Ca²⁺ (Fig. 11 C), further reflecting the reduced affinity for Ca²⁺ in the LCa_v3 pore. Collectively, the Ca²⁺ block data indicates that TCa_v3 exhibits a potent, but perhaps slightly diminished Ca²⁺-block of Na⁺ currents compared with human Ca_v3.1. Instead, the two exon 12 variants of Lymnaea Ca_v3 exhibit moderately attenuated and extremely attenuated Ca²⁺ block, indicative of lowered binding affinity for Ca²⁺ along the extracellular surface of the channel pore.

To further understand the Na⁺ permeation properties of TCa_v3, we characterized its Ca²⁺ versus monovalent cation selectivity by measuring zero-current reversal potentials (E_Rev) under bi-ionic conditions (i.e., 4 mM Ca²⁺_out and 100 mM Li⁺_in, Na⁺_in, K⁺_in, or Cs⁺_in). The reversal potential of peak bi-ionic currents is determined by the pore’s preference for inward-permeating Ca²⁺, which pulls E_Rev toward more positive voltages, versus outward-permeating monovalent cations (i.e., X⁺, where X = Li, Na, K, or Cs), which pull E_Rev toward more negative voltages. Depolarizing voltage steps from −90 to 70 mV under the four bi-ionic conditions (i.e., Ca²⁺_out-Li⁺_in, Ca²⁺_out-Na⁺_in, Ca²⁺_out-K⁺_in, and Ca²⁺_out-Cs⁺_in) produce markedly different outward current components for TCa_v3, with decreasing amplitudes from Li⁺ to Na⁺ to K⁺ to Cs⁺ (Fig. 12 A), reflecting the pore’s decreasing permeability to monovalent cation flow according to the Eisenman selectivity model (i.e., ions with smaller radii are more permeable through a narrow pore: Li⁺ < Na⁺ < K⁺ < Cs⁺ with respect to radius; Eisenman et al., 1967; Eisenman and Horn, 1983). Correspondingly, E_Rev values for the four bi-ionic IV plots for TCa_v3 have leftward shifts corresponding to increased monovalent permeation from Cs⁺ to Li⁺ (Fig. 12 B). Converting E_Rev values to permeability ratios P_Ca/P_X using the bi-ionic Nernst equation (Senatore et al., 2014), which reflect the pore’s preference for Ca²⁺ over monovalent X⁺, reveals that the Trichoplax channel is poorly selective for Ca²⁺ over X⁺ monovalents compared with human Ca_v3.1 (Stephens et al., 2015), falling between previously published values for Lymnaea LCa_v3-12a and LCa_v3-12b variants for P_Ca/P_Na and P_Ca/P_K (Fig. 12 C; Senatore et al., 2014). Interestingly, TCa_v3 is the most Li⁺-permeable channel, whereas the Cs⁺ permeability for all non-deuterostome channels is roughly equal and significantly higher than human Ca_v3.1.

Ca²⁺ versus X⁺ permeability features of T-type channels are also reflected in the rectification of macroscopic bi-ionic currents, where three conductance states are evident on IV plots: (1) an inward conductance at voltages where the driving force favors inward flow of Ca²⁺ (G_In); (2) a conductance through E_Rev where a transition between inward Ca²⁺ and outward X⁺ occurs, (G_Rev); and (3) an outward conductance where driving force favors outward X⁺ flow (G_Out; Fig. 12 D and Fig. S3). Comparison of the three conductance states of TCa_v3 with previously published data for other in vitro–expressed T-type channels (Senatore et al., 2014; Stephens et al., 2015) corroborates a low Ca²⁺ selectivity for the Trichoplax Ca_v3 channel compared with human Ca_v3.1. Both G_Rev and G_Out values for TCa_v3, which reflect permeability to monovalent cations, are significantly larger than those of Ca_v3.1, regardless of monovalent cation type (Fig. 12 E). Indeed, Ca_v3.1 is very ineffective at conducting outward monovalent currents, even at extremely depolarized potentials >60 mV (Fig. 12 D and Fig. S3). Ca_v3.1 also has the largest conductance for inward Ca²⁺ flow (G_In), but noticeably, under the most physiological conditions with K⁺ present in the internal saline, the G_In Ca²⁺ conductance for TCa_v3 encroaches on that of Ca_v3.1 (i.e., means are not statistically different; Fig. 12 E). At the other extreme, the Lymnaea channel LCa_v3-12a has the lowest conductance for inward Ca²⁺ (G_In) and the largest monovalent conductances through E_Rev and outwards (G_Rev and G_Out, respectively; Fig. 12 D and Fig. S3). Collectively, the bi-ionic reversal potential data indicate that the TCa_v3 channel pore is poorly selective for Ca²⁺ over monovalents compared with human Ca_v3.1, with bi-ionic selectivity properties more similar to those of highly Na⁺-permeant LCa_v3-12a and the more Ca²⁺-selective LCa_v3-12b channels (Table 2).

In light of these data, it is apparent that Ca²⁺ block, and not Ca²⁺ selectivity, is the major determinant for the degree of Na⁺ that permeates through TCa_v3 and other T-type channels. Indeed, the efficiency of Ca²⁺ block for the various cloned T-type channels (Ca_v3.1 > TCa_v3 > LCa_v3-12b > LCa_v3-12a; Fig. 11) correlates with their degree of Na⁺ permeation (e.g., 21.0%, 29.6%, 60.5%, and 93.5% of total current carried by Na⁺, respectively; measured by dividing percentage increases in peak current shown in Fig. 10 B by that same value plus 100%).

Ca_v3 channels appear to have emerged upwards of a billion years ago in a eukaryotic ancestor of choanoflagellates and metazoans (Morris, 1998), being present in the genome of S. rosetta (Fairclough et al., 2013; Moran and Zakon, 2014). We were unable to find a Ca_v3 channel homologue in the genome of choanoflagellate Monosiga brevicollis (King et al., 2008), indicative of either gene loss or an incomplete genome sequence. We also failed to identify Ca_v3 channel genes in various genomes and transcriptomes available for sponges and ctenophores (Srivastava et al., 2010; Ryan et al., 2013; Moroz et al., 2014; Fernandez-Valverde et al., 2015), suggesting that these two phyla lost Ca_v3 channels.

Analysis of various Ca_v3 channel gene exon/intron structures revealed a general trend toward increased intron/exon number from choanoflagellates to vertebrates, with the Salpingoeca Ca_v3 gene bearing 13 exons from start codon to stop codon, Trichoplax Ca_v3 bearing 28 exons, Nematostella Ca_v3a bearing 30 exons, and mouse Cav3.1 bearing 38 exons (Fig. 3). Such an increase in intron number from premetazoans, to early diverging metazoans, to bilaterians is consistent with observed trends at the whole-genome level (King et al., 2008). Interestingly, Ca_v3 channel genes from bilaterians appear to have undergone the most significant changes in intron/exon structure within coding sequences for channel structures associated with modulation of function, such as optional exon 8b found in mouse Ca_v3.1 and Lymnaea Ca_v3, which regulates channel membrane expression (Senatore and Spafford, 2012); mouse exons 14–17 in the II–III linker, which serve as a hotbed for modulation by kinases and G-proteins (Chemin et al., 2006; Perez-Reyes, 2010b; Senatore et al., 2012); exons 25c and 26 in the III–IV linker, which in both vertebrates and protostome invertebrates cause alterations in channel voltage-gating and kinetics (Senatore and Spafford, 2012); the C terminus, where vertebrate Ca_v3 channels physically interact with other ion channel types such as Ca²⁺-modulated A-type K⁺ channels (Anderson et al., 2010); and SNARE proteins, which facilitate T-type channel involvement in low-threshold exocytosis (Weiss et al., 2012). Indeed, if a similar pattern extends to numerous other gene orthologues in the metazoan genome, it might account for some of the apparent increase in anatomical, cellular, and molecular complexity of vertebrates and bilaterians compared with more early diverging animals (Valentine et al., 1994), in spite of comparable total gene counts.

Despite its comparative simplicity at the gene structure level, the TCa_v3 protein sequence retains all of the hallmark structural features of four-domain channels, with four homologous repeat domains, each containing extracellular turret-containing P-loops separated by pore-forming S5 and S6 helices, and voltage-sensor modules (S1 to S4 helices), with S4 helices packed with positively charged lysine (K) and arginine (R) residues critical for voltage sensing (Fig. 1 A; Catterall, 2010). In addition, TCa_v3 bears the three most prominent structural features that distinguish Ca_v3 channels from Ca_v1 and Ca_v2 types: (1) a selectivity filter motif with two aspartates (D) instead of two glutamates (E) in the P-loops of domains III and IV (i.e., EEDD vs. EEEE); (2) the absence of a calmodulin-binding isoleucine-glutamine (IQ) motif in the C terminus, which permits modulation of Ca_v1 and Ca_v2 channels by cytoplasmic Ca²⁺ influx (Ben-Johny et al., 2014); and (3) a predicted helix-turn-helix gating brake structure in the domain I–II intracellular linker, in an analogous region where Ca_v1 and Ca_v2 channels bind accessory Ca_vβ subunits via a distinct structure called the α interaction domain (AID; Perez-Reyes, 2010a).The absence of an AID in T-type channels highlights another distinguishing feature, which is a lack of dependence on Ca_vβ subunits. For Ca_v1 and Ca_v2 channels, Ca_vβ as well as Ca_vα₂δ subunits are obligate counterparts that complex with the channels to regulate gating, trafficking, and proteolytic turnover (Arikkath and Campbell, 2003; Richards et al., 2004; Altier et al., 2011). Instead, T-type channels function as separate entities, with an autonomous gating brake serving in lieu of the AID/Ca_vβ subunit, which nevertheless also regulates channel gating (Perez-Reyes, 2010a). Even the most basal of all known T-type channels, from S. rosetta, bears a predicted gating brake motif in its I–II linker (Fig. 2 D). The structural distinction between AIDs of high voltage–activated Ca_v1/Ca_v2 channels and gating brakes of low voltage–activated Ca_v3 channels thus appears to have ancient origins predating Metazoa. However, given the similar helical arrangement that extends into the cytoplasm from the domain I S6 helix, it is conceivable that both structures evolved via divergence from a helical structure present in an ancient Ca_v channel ancestor of Ca_v1/Ca_v2 and Ca_v3 channels.

In bilaterians, the three types of Ca_v channels are specialized to carry out distinct and pivotal roles in neurons and muscle, and other excitable cell types, where they translate electrical signals of Na_v, K_v, and synaptic ligand-gated cation channels into cellular events by coupling with Ca²⁺-sensitive cytoplasmic proteins (Berridge, 2006; Rizzuto and Pozzan, 2006). Ca_v1 channels are most classically associated with excitation-contraction coupling in muscle, as well as excitation-transcription coupling in neurons and muscle, and Ca_v2 channels are associated with fast presynaptic exocytosis of neurotransmitters (excitation-secretion coupling; Catterall, 2011). Instead, Ca_v3 channels have long eluded such a stereotyped and ubiquitous classification, where their roles tend to vary depending on cell type, ranging from regulating cellular excitability in select neurons and other excitable cells, to driving low-threshold exocytosis in select neurons and neurosecretory cells, to regulating tone and contraction of various muscle cell types (Perez-Reyes, 2003; Senatore et al., 2012).

Trichoplax is very interesting in its highly simplified cellular body plan, bearing only six cell types: ciliated dorsal and ventral epithelial cells, with ventral epithelial cells responsible for ciliary locomotion (i.e., gliding along hard surfaces); crystal cells, which bear internal birefringent crystals with unknown function (Smith et al., 2014); fiber cells, positioned between the epithelial cell layers and proposed to be contractile in nature (Behrendt and Ruthmann, 1986; Smith et al., 2014); gland cells, proposed to play roles in paracrine signaling and which resemble neurons and neurosecretory cells in their expression of exocytotic SNARE proteins and membrane-apposed vesicles (Grell and Ruthmann, 1991; Smith et al., 2014); and ventral lipophil cells, also apparently exocytotic in nature but specialized for secretion of hydrolytic enzymes for external digestion of algae during feeding (Smith et al., 2014, 2015). The apparent morphological, anatomical, and ultrastructural similarly of Trichoplax cell types with those from more complex animals (i.e., epithelial, neuron/neuroendocrine, muscle, and digestive) is suggestive of cellular homology at the level of genes and proteins. In accordance, the presence of numerous genes in the Trichoplax genome crucial for cell-specific functioning (Srivastava et al., 2008), including those for generating and packaging neurotransmitters and neuropeptides (Nikitin, 2015), cellular excitability (e.g., Na_v2 and Kv channels, K⁺ leak channels; Senatore et al., 2016), and muscle contraction (Steinmetz et al., 2012), suggests that homologous molecular processes are taking place in some of these cell types.

Here, we extend the apparent homology of gland cells with neurons and neuroendocrine cells in our localization of TCa_v3 to this cell type (Fig. 4), suggesting that the cells undergo rapid fluxes in membrane voltage and transient rises in cytoplasmic [Ca²⁺], perhaps for engaging Ca²⁺ sensitive elements of the exocytotic machinery. A Ca²⁺ dependence for gland cell exocytosis is further suggested by their expression of complexin, a Ca²⁺-responsive regulator of the SNARE complex and vesicle fusion (Yoon et al., 2008), and further, by the presence of synaptotagmin in the genome (Srivastava et al., 2008), a key Ca²⁺ sensor for exocytotic release which interacts with complexin during fusion (Tang et al., 2006). We note from our RT-PCR experiment that beyond the Ca_v3 channel, Trichoplax expresses Ca_v1 and Ca_v2 channels at the whole-animal mRNA level, as well as the single Ca_vβ and three Ca_vα₂δ accessory subunit genes (Fig. 5). If the Ca_v1 and/or Ca_v2 channels are similarly localized to gland cells, TCa_v3 might thus serve to activate the high voltage–activated channels that in turn drive vesicle fusion and exocytosis, as is common in neuroendocrine cells (Mansvelder and Kits, 2000).

Notably, the apparent enrichment of TCa_v3 along the outward-facing area of gland cells (Fig. 4), where the SNARE proteins and membrane-apposed vesicles were previously shown to reside (Smith et al., 2014), is suggestive of a more direct involvement in exocytosis, perhaps driving graded exocytosis before action potential threshold as occurs in select vertebrate and invertebrate neurons and neurosecretory cells (Carbone et al., 2006; Weiss and Zamponi, 2013; Senatore et al., 2016). In vertebrates, a direct interaction between Ca_v3 channels and SNARE proteins syntaxin-1A and SNAP-25 has been documented (Weiss et al., 2012), tethering the channels close to Ca²⁺-sensitive elements of the exocytotic machinery. The expression of TCa_v3 in gland cells implies that the use of T-type channels in regulating excitation and secretion of neuron/neuroendocrine-like cells might have evolved very early on, at least in the ancestor of placozoans and cnidarians/bilaterians, but perhaps even further back, in the single-celled ancestor of choanoflagellates and metazoans, which possessed a T-type channel gene as well as the core elements of the exocytotic apparatus (King et al., 2008; Fairclough et al., 2013; Moran and Zakon, 2014).

Our first efforts to express the cloned TCa_v3 channel in HEK-293T cells for electrophysiological recording were unsuccessful. Instead, we found that coexpression with rat Ca_vβ₁b and Ca_vα₂δ₁ accessory subunits of high voltage–activated Ca_v1 and Ca_v2 channels dramatically increased channel expression either as a fusion protein with EGFP, or as a separate protein recordable via whole-cell patch clamp. Interestingly, Ca_vβ and Ca_vα₂δ subunits increase membrane expression of high voltage-activated Ca_v channels in part by blocking their internalization and proteasomal degradation (Bernstein and Jones, 2007; Altier et al., 2011; Dolphin, 2012), although the process seems to depend on direct interactions between the subunits and the channel proteins. We do not necessarily expect direct protein–protein interaction between TCa_v3 and the Ca_vβ₁b and Ca_vα₂δ₁ accessory subunits because such an interaction has yet to be reported for any other cloned vertebrate or invertebrate T-type channel (Dubel et al., 2004; Senatore and Spafford, 2010; Dawson et al., 2014; Cens et al., 2015; Jeong et al., 2015). However, given the overlapping emergence of Ca_v1/Ca_v2 channels, Ca_v3 channels, and the Ca_vβ subunit in a premetazoan ancestor (Dawson et al., 2014; Moran and Zakon, 2014), the possibility exists that ancestral T-type channels and those from extant basal organisms such as Trichoplax physically interact with high voltage-activated Ca_vβ₁b and Ca_vα₂δ₁ subunits. Here, we found that rat Ca_vβ₁b and Ca_vα₂δ₁ subunits significantly increased ectopic expression of GFP expressed in the absence of TCa_v3 (Fig. 6), indicating that their effect on ectopic protein expression might at least in part be a result of nonspecific processes.

One of the most clear cellular functions of Ca_v3 channels is regulating excitability (Perez-Reyes, 2003; Senatore et al., 2012), invoking their low voltages of activation and fast, transient kinetics. For example, mammalian Ca_v3.2 calcium channels are enriched in pain-sensing neurons, where they amplify depolarizing sensory inputs to increase nociceptive signaling to the spinal cord (Bourinet et al., 2005; Rose et al., 2013). Similarly, in the brain, neuronal Ca_v3 channels are enriched along dendrites (McKay et al., 2006), where they boost postsynaptic excitatory potentials to increase the likelihood of eliciting action potentials (Perez-Reyes, 2003; Senatore et al., 2012). In some cases, the rapid kinetics and low voltages of activation of Ca_v3 channels enable them to function in lieu of Na_v channels where they can drive Ca²⁺ action potentials, as occurs in striated muscle cells from jellyfish (Lin and Spencer, 2001), snail cardiomyocytes (Yeoman et al., 1999; Senatore et al., 2014), and C. elegans pharyngeal muscle (Steger et al., 2005).

Here, we show that the biophysical properties of the Trichoplax Ca_v3 channel are consistent with a role in regulating excitability in gland cells. Inward Ca²⁺ currents recorded from HEK cells expressing recombinant TCa_v3 emerge upon slight membrane depolarization from a holding potential of −110 mV (Fig. 7, A and B), indicating that like all other in vitro–expressed Ca_v3 channels, TCa_v3 is low voltage activated. In fact, TCa_v3 has the lowest voltage dependency for activation of any cloned T-type channel, with a 5 mV more negative maximal peak inward current than Lymnaea Ca_v3, and 10 to 20 mV more negative than the three mammalian isotypes, Ca_v3.1 to Ca_v3.3 (Fig. 7 B and Table 2). Accordingly, Boltzmann transformation of the peak current–voltage (IV) plot for TCa_v3, into an activation curve (Fig. 7 C), reveals a half-maximal activation considerably left-shifted compared with other Ca_v3 channels. Instead, the channel’s half-maximal steady-state inactivation is roughly similar to those of other channels (Table 2). TCa_v3 currents also have reasonably fast activation and inactivation kinetics, which although marginally slower than Lymnaea Ca_v3 and the mammalian Ca_v3.1/Ca_v3.2 isotypes, are considerably faster than those of Ca_v3.3, the slowest of the vertebrate T-type channels (Table 2). TCa_v3 is thus capable of conducting fast inward Ca²⁺ currents upon slight membrane depolarization, with a lower voltage threshold than other T-type channels, while being equally subject to voltage-dependent inactivation. These features indicate that TCa_v3 is poised to be more active at threshold voltages compared with other T-type channels. Furthermore, the particularly slow deactivation kinetics for TCa_v3, at voltages near −70 mV (Table 2), would serve to counter K_v channel–driven action potential repolarization, effectively widening action potentials and increasing net Ca²⁺ influx. Overall, the voltage dependencies and kinetic properties of TCa_v3 are most similar to those of fellow invertebrate T-type, Lymnaea Ca_v3, and least to those of mammalian Ca_v3.3 (Table 2). What is striking is that upon side-by-side comparison of biophysical properties of various cloned Ca_v3 channels, the structurally divergent Trichoplax homologue, which is >600 million years separated from mammalian isotypes, is more similar to mammalian Ca_v3.1 and Ca_v3.2 than is Ca_v3.3. Indeed, there appear to have been strong evolutionary constraints on the TCa_v3 channel to retain a core set of biophysical properties, suggesting that the need for its cellular contributions are conserved even in the absence of neurons and muscle.

An important caveat toward speculation about the physiological roles for TCa_v3 in vivo is that its contributions would ultimately depend on the membrane potential, which is controlled by a milieu of electrogenic proteins. Although Trichoplax has the majority of these electrogenic genes (Srivastava et al., 2008; Senatore et al., 2016), suggesting that some of its cells have polarized resting membrane potentials and exhibit rapid fluxes in membrane voltage such as action potentials, the membrane properties of Trichoplax cells have yet to be reported. We and others have attempted intracellular recording of isolated Trichoplax cells, but their small size (<10-µm diameter) and particular membrane features make obtaining a gigaohm seal during patch-clamp recording particularly difficult. Based on the biophysical properties of TCa_v3, we can speculate that if the resting membrane potential of a typical Trichoplax cell sits above −60 mV, the channel would not contribute to excitability because of inactivation (Fig. 7 C). However, transient hyperpolarization from such potentials could recruit the channel by removing inactivation, where it would contribute to postinhibitory rebound (PIR) excitation. In mammals, T-type channel-mediated PIR excitation plays an important role in certain neuronal circuits, such as the thalamus, where postinhibitory Ca²⁺ spikes support rhythmic bursts of action potentials that project to the cortex and gate sensory information during non-REM sleep (Lee et al., 2004; Anderson et al., 2005; Crunelli et al., 2006). Notably, TCa_v3 and the Ca_v3 channel from Lymnaea have a slower recovery from inactivation than mammalian channels (Table 2), so they would require more prolonged hyperpolarization to be recruited for PIR excitation. Finally, even if Trichoplax cells do not undergo rapid changes in membrane voltage, T-type channels could nevertheless contribute a consistent steam of Ca²⁺ into the cytosol through a window current (Fig. 7 C), which can be used by cells to transition between bimodal resting membrane potentials (Dreyfus et al., 2010) and are associated with cellular proliferation during development and cancer (Lory et al., 2006; Senatore et al., 2012; Gackière et al., 2013).

Previous work examining altered Na⁺ permeation of the Lymnaea T-type channel, caused by alternative splicing of exons 12a and 12b in a region of the domain II P-loop called the turret, revealed that factors outside of the selectivity filter can nevertheless have important consequences for defining Ca²⁺ versus Na⁺ permeation. However, based on these studies, it is difficult to reconcile differences in permeation properties among Ca_v3 channels purely on the structure of exon 12, where in Lymnaea, the smaller exon 12a imposes extreme Na⁺ permeability (i.e., 93.5% of current carried by Na⁺; Fig. 10 B: 1,440 ÷ 1,540% = 93.5%), whereas in mammalian channels, homologous 12a-like exons produce only moderate Na⁺ permeability (∼21.0%, 24.3%, and 31.3% for Ca_v3.1, Ca_v3.2, and Ca_v3.3, respectively). We found the basal TCa_v3 channel to resemble mammalian channels, with ∼29.6% of inward current carried by Na⁺, making it statistically indistinguishable from Ca_v3.2 and Ca_v3.3 (Table 2). Indeed, a low Na⁺ permeability for T-type channels bearing exon 12a–like turrets extends to other non-protostome channels that span the lineages between placozoans and mammals, including those from cnidarians (i.e., jellyfish; Lin and Spencer, 2001) and deuterostomes (i.e., echinoderm starfish eggs; Hagiwara et al., 1975), where T-type currents were reported to be carried mostly by Ca²⁺. Based on the available data, one explanation for the emergence of altered Na⁺ permeation in the snail and other protostome T-type channels via exon 12 splicing is that after dupliction of exon 12, structural alterations took place outside of the exon 12 region, rendering 12a-bearing channels more Na⁺ permeable. The duplicated exon 12b, unique to protostomes, was possibly adaptated to retain Ca²⁺ conducting channels via enlargement relative to exon 12a by ∼11 aa and increase in cysteine content from 0–3 to ∼5. In Lymnaea, exon 12b produces a channel more in line with non-protostome 12a-like channels, from Trichoplax through to mammals, with only ∼60.5% of current carried by Na⁺.

Interestingly, the degree of co-permeation of Na⁺ alongside Ca²⁺ for the various cloned T-type channels seems to hold true even in the presence of Ba²⁺, which can exhibit increased or decreased permeation relative to Ca²⁺ in a channel-dependent manner. Like rat Ca_v3.1, TCa_v3 conducts larger Ca²⁺ currents than Ba²⁺ currents in vitro, whereas rat Ca_v3.2, LCa_v3-12a, and LCa_v3-12b all conduct larger Ba²⁺ currents and rat Ca_v3.3 conducts equal Ca²⁺ and Ba²⁺ currents (McRory et al., 2001). Despite these differences, the pattern of fold increases in peak inward current for the various channels upon replacement of impermeant external NMDG⁺ with Na⁺ is consistent regardless of whether Ca²⁺ or Ba²⁺ is present in the extracellular solution (Fig. 10). As such, it appears as though the factors that determine macroscopic conduction preference among divalent cations are different from those that determine preference between divalent versus monovalent cations.

Notably, a previous study found that NMDG⁺ might directly block inward current through T-type channels (Khan et al., 2008), potentially confounding loss of current amplitude caused by replacement of Na⁺ in our experiments. However, for the snail T-type channel exon 12a and 12b variants, we previously found that replacement of Na⁺ with impermeant Tris⁺ resulted in a similar pattern of current attenuation compared with NMDG⁺ (Senatore, 2012), suggesting that the major effect on T-type channel current amplitude in these experiments is caused by Na⁺ depletion, and not NMDG⁺ block.

We sought to identify different aspects of cation permeation through T-type channels that could account for their varying Na⁺ permeabilities. For high voltage–activated Ca_v1 and Ca_v2 channels, which are considerably better than T-types at selecting for Ca²⁺ (e.g., compare 21–90% Na⁺ current for Ca_v3 channels with <0.1% for Ca_v1/Ca_v2; Tsien et al., 1987; Sather and McCleskey, 2003; Shcheglovitov et al., 2007; Cheng et al., 2010; Buraei et al., 2014; Tang et al., 2014), selectivity for Ca²⁺ is attributed to the ability of the ion to associate with a high-affinity binding site at the extracellular surface of the pore to repel and block inward Na⁺ flux (i.e., Ca²⁺ block). T-type channels are also expected to bind Ca²⁺, but their reduced selectivity is attributed to a ∼10-fold lowered binding affinity and reduced Ca²⁺ block (Shcheglovitov et al., 2007). We compared the Ca²⁺ block properties of various Ca_v3 channels, revealing that TCa_v3 shares with mammalian channels a potent Ca²⁺ block, indicated by rapidly attenuating Na⁺ currents as Ca²⁺ is incrementally added to the extracellular solution (Fig. 11). By stark contrast, exon 12 variants of LCa_v3 exhibit a reduction in Ca²⁺ block that is most extreme for LCa_v3-12a, for which the Ca²⁺ titration curve does not exhibit the classic U shape (where decreasing Ca²⁺-blocked Na⁺ currents gradually give way to increasing Ca²⁺ currents as [Ca²⁺]_out increases; Fig. 11 A). Instead, the channels exhibit continued decline through to 10 mM [Ca²⁺]_out, reflecting dramatically reduced Ca²⁺ binding affinity in the pore. Altogether, the Ca²⁺ block properties of the different Ca_v3 channels (Fig. 11) correlate with their respective Na⁺ permeabilities, apparent in Na⁺/NMDG⁺ replacement experiments (Fig. 10 B): TCa_v3 and human Ca_v3.1 have the most potent Ca²⁺ block and lowest Na⁺ permeability, and LCa_v3-12a and LCa_v-12b have reduced Ca²⁺ block proportional to their respective increases in Na⁺ permeability.

However, we point out a minor inconsistency between our Ca²⁺-Na⁺ permeation data (Fig. 10) and our Ca²⁺ block data (Fig. 11). Whereas the Ca²⁺ block data suggests that at near-physiological external Ca²⁺ concentrations (i.e., 1 mM [Ca²⁺]_out), most of the Na⁺ current has been blocked for all T-type channels (Fig. 11, A and B), replacement of 135 mM NMDG⁺ with Na⁺ in the presence of 2 mM [Ca²⁺]_out causes significant increases in peak inward current (most marked for LCa_v3-12a), reflecting additive Na⁺ currents over already present Ca²⁺ currents (Fig. 10, A and B). So a question arises: why in 2 mM Ca²⁺ can you observe a considerable additive Na⁺ current via NMDG⁺ replacement, whereas 1–3 mM Ca²⁺ seems to mostly block the Na⁺ currents? We explain this inconsistency by noting that in Ca²⁺ block experiments, there are considerable residual currents, ranging in amplitudes from 100–200 pA (i.e., for LCa_v3-12b) to 100–800 pA (for LCa_v3-12a), which persist despite saturation in the Ca²⁺ block effect. We suggest that these residual currents represent combined Ca²⁺/Na⁺ currents, which would be differentially attenuated if the Na⁺ ions were to be replaced with impermeant NMDG⁺, consistent with the Na⁺/NMDG⁺ substitution data (Fig. 10, A, B, E, and F).

TCa_v3 appears to be slightly more Na⁺ permeable than human Ca_v3.1, in part because of a slightly reduced Ca²⁺-block and pore Ca²⁺-binding affinity (Fig. 11, A and B). However, other aspects of the pore might influence the degree of Na⁺ permeation. Our bi-ionic reversal potential experiments, which approximate the preference of a given channel pore for inward-flowing Ca²⁺ ions versus outward-flowing monovalents (i.e., Li⁺, Na⁺, K⁺, and Cs⁺), revealed that TCa_v3 was considerably less selective for Ca²⁺ than human Ca_v3.1 (e.g., P_Ca/P_Na = 35.61 ± 1.52 vs. 89.56 ± 8.21, respectively), falling between the more Na⁺ permeable channels LCa_v3-12a (33.06 ± 1.50) and LCa_v3-12b (41.49 ± 1.98). Indeed, if preference for Ca²⁺ over monovalents measured under bi-ionic conditions was the major determinant for Na⁺ permeation, then we would expect TCa_v3 to conduct a much higher proportion of inward Na⁺ upon depolarization, most similar to the LCa_v3-12a variant (Fig. 10 B). Instead, the selectivity for Ca²⁺ over monovalents in the TCa_v3 channel pore seems somewhat inconsequential, where 29.6% of the total current is carried by Na⁺, which is more similar to Ca_v3.1 (21.0%) than either LCa_v3-12b (60.5%) or LCa_v3-12a (93.5%). As such, Ca²⁺ block appears to be more consequential for determining Na⁺ permeation than is Ca²⁺ over Na⁺ selectivity. Instead, Ca²⁺ selectivity appears to play a more marginal role, possibly accounting for the 8.6% increase in Na⁺ permeability for TCa_v3 relative to Ca_v3.1.

Finally, we also find an interesting inverse correlation between conductance values across bi-ionic reversal potential (G_Rev) and outward monovalent current flow (G_Out) versus the potency of the Ca²⁺-block effect: whereas LCa_v3-12a has the largest G_Rev and G_Out values and the weakest Ca²⁺ block, Ca_v3.1 has the smallest G_Rev and G_Out values and the strongest Ca²⁺ block (compare Fig. 11 B with the top two panels of Fig. 12 E). This is not surprising because G_Rev and G_Out should reflect the pore’s permeability to outward-flowing monovalents, against external Ca²⁺ which seeks to bind the extracellular high-affinity site. However, there is a slight inconsistency in this correlation, where G_Rev and G_Out conductance values for TCa_v3 are significantly larger than Ca_v3.1, despite both channels bearing similarly potent Ca²⁺ block properties. Instead, the conductance values for TCa_v3 lie between those of Ca_v3.1 and the two exon 12 splice variants of LCa_v3. Thus, conductance across the reversal potential and during outward current flow though T-type channels might depend on a combination of both affinity for Ca²⁺ at the external pore surface (where TCa_v3 and Ca_v3.1 are similar), and pore Ca²⁺ versus Na⁺ permeability (where TCa_v3 and LCa_v3 are more similar).

In conclusion, we find evidence that for T-type channels, Ca²⁺ block—and not Ca²⁺ versus Na⁺ selectivity—best correlates with the degree of inward Na⁺ permeation under simulated physiological conditions. For Trichoplax, which lives in seawater, the abundance of external Ca²⁺ must ensure efficient saturation of the Ca²⁺ block effect, rendering the channels mostly permeable to Ca²⁺. Here, our studies were performed using salines that are compatible with HEK-293T cells, with reduced ionic concentrations and osmolarity across both sides of the membrane compared with seawater. In future studies, it will be interesting to evaluate the consequences of altered ion concentrations on TCa_v3 biophysical, permeation, and pharmacological properties (and indeed other ion channel types), where evolutionary transitions from seawater to land/freshwater environments, and perhaps back again, would likely require some level of adaptation in channel function, such that contributions to cellular excitability and Ca²⁺ influx remain within acceptable parameters.

We thank Drs. David Spafford and Paul S. Katz for providing support to A. Senatore during the preliminary stages of this research, Dr. Arnaud Monteil for preliminary Western blotting of the TCa_v3 channel protein, Farid R. Ahmadli for help analyzing Ca_v3 channel gene sequences, and Dr. Andreas Heyland for providing the Trichoplax specimens used for sequencing and cloning of the TCa_v3 channel cDNA.

This work was funded by a National Science and Engineering Research Council of Canada (NSERC) Discovery Grant (RGPIN-2016-06023), a Canadian Foundation for Innovation Grant (CFI Project 35297), University of Toronto Mississauga start-up funds, and an NSERC postdoctoral fellowship (PDF-43851-2013) to A. Senatore.

The authors declare no competing financial interests.

Richard W. Aldrich served as editor.