In the pre-synaptic terminal of bilaterians, Ca_v2 channels play a dominant role in coupling excitation with fast pre-synaptic exocytosis. Ca_v1 and/or Ca_v3 channels provide more peripheral contributions, or contribute to other forms of excitation-secretion coupling such as neuroendocrine secretion (Ca_v1 and Ca_v3 channels) and low-threshold exocytosis (Ca_v3 channels) (Carbone et al., 2006a; Simms and Zamponi, 2014). One requirement for fast synaptic transmission is the proximal coupling of pre-synaptic Ca_v2 channels with Ca²⁺-sensitive proteins of the exocytotic machinery (e.g., synaptotagmin, complexin), such that transient “nanodomain” Ca²⁺ plumes, restricted to roughly 100 nanometer radii from the channel pore, effectively saturate the exocytotic apparatus (Clapham, 2007; Stanley, 2016). Two modes for functional coupling of Ca_v2 channels with exocytotic proteins appear to exist: (1) In nanodomain coupling, direct, physical interaction of Ca_v2 channels with proteinaceous elements of docked pre-synaptic vesicles allows single channels to trigger exocytosis of single vesicles (i.e., one-to-one coupling); (2) Instead, microdomain coupling involves a slightly more distal apposition between Ca_v channels and synaptic vesicles, without necessarily direct physical contact. Here, numerous Ca_v channel Ca²⁺ nanodomains sum into larger “microdomains,” which trigger exocytosis of numerous docked vesicles (i.e., group-to-group coupling) (Stanley, 2016).

Fast chemical synaptic transmission in cnidarians, as in bilaterians, requires Ca²⁺ influx through voltage-gated calcium channels (Bullock, 1943; Kerfoot et al., 1985). However, whether cnidarian Ca_v2 channels similarly act as major drivers of pre-synaptic exocytosis remains to be determined. A recent study of the spatial expression of Ca_v channel mRNAs in the developing anthozoan sea anemone N. vectensis (Moran and Zakon, 2014), revealed that all of its Ca_v channel genes, including its three Ca_v2 channels, are expressed in regions that overlap with expressed neurogenic marker genes ELAV and Musashi identified in a separate study (Marlow et al., 2009). However, direct evidence that Ca_v2 channels are expressed in cnidarian neurons and exhibit pre-synaptic localization has yet to be provided. Indirectly, electrophysiological recordings of motor neurons from the hydrozoan jellyfish Polyorchis penicillatus reveal a prominent HVA Ca²⁺ current that resembles Ca_v2 channels by lacking fast Ca²⁺/calmodulin-dependent inactivation (Przysiezniak and Spencer, 1992), a conserved feature of protostome and deuterostome Ca_v1-type channels (Peterson et al., 1999; Spafford et al., 2006; Taiakina et al., 2013), also evident for the cloned and ectopically expressed Ca_v1 channel from jellyfish Cyanea capillata (Jeziorski et al., 1998). Interestingly, neuromuscular junction (NMJ) synapses in Polyorchis have properties suggestive of nanodomain (one-to-one) coupling. In vitro voltage-clamp recording across the NMJ revealed that rapid and transient pre-synaptic Ca²⁺ influx elicits stronger post-synaptic responses than slower, more long-lasting Ca²⁺ influx, despite the former providing less total Ca²⁺ into the pre-synaptic terminal (Spencer et al., 1989). The increased efficiency for synaptic transmission with faster onset Ca²⁺ influx suggests that endogenous mechanisms for Ca²⁺ sequestration and extrusion impose spatial/temporal constraints on the calcium channel's ability to activate the exocytotic machinery. Instead, efficient transmission at this particular synapse requires fast plumes of cytoplasmic Ca²⁺, consistent with nanodomain coupling. Synapses with microdomain coupling tend to improve their efficacy (i.e., facilitate) with increased pre-synaptic Ca²⁺ influx, such as occurs during a burst of action potentials (Stanley, 2016). Such Ca²⁺-dependent facilitation has been observed in other cnidarian synapses (Roberts and Mackie, 1980), suggesting they are similar to vertebrates in having different synapses with either nanodomain or microdomain coupling, depending on developmental state or physiological requirements (Stanley, 2016).

Given the similar genomic content of pre-and post-synaptic genes between bilaterians and cnidarians, and their shared ancestry of the nervous system, it will be interesting to evaluate the homology in their mechanisms for synaptic transmission at the molecular level. With respect to nanodomain tethering of Ca_v2 channels, there is at least evidence for homology between protostomes and deuterostomes, which diverged roughly 520 Mya (Blair and Hedges, 2005), around when medusozoans and anthozoans diverged from each other (Putnam et al., 2007). In both Drosophila and mouse, the presynaptic scaffolding protein Rab-3 interacting molecule (RIM) is essential for the appropriate pre-synaptic localization of Ca_v2 channels, forming part of a molecular bridge between the channels and synaptic vesicles (Han et al., 2011; Kaeser et al., 2011; Graf et al., 2012). The interaction seems to occur via a RIM PDZ motif that binds the Ca_v2 channel C-terminus; in mouse, targeted deletion of the RIM PDZ disrupts proper channel localization and synaptic transmission, and a direct physical interaction was observed between this motif and the channel C-terminus via yeast-two hybrid and NMR spectroscopy assays. However, a similar interaction was not observed in the chick synapse using co-immunoprecipitation (Khanna et al., 2006; Wong and Stanley, 2010), and a separate study found that RIM tethering of Ca_v2 channels requires the Ca_vβ subunit to serve as an intermediary between the two proteins (Kiyonaka et al., 2007). Recently, evidence has emerged that Ca_v2 channel pre-synaptic scaffolding undergoes a developmental switch in Drosophila, where different mRNA splice isoforms of the vesicle priming protein UNC-13 interact with distinct scaffolding proteins for either microdomain tethering in immature synapses (i.e., with Syd-1 and Liprin-α), or nanodomain tethering in mature synapses (i.e., with Bruchpilot and RIM-associated protein complexes) (Böhme et al., 2016). Thus, although it appears as though RIM plays conserved roles in Ca_v2 channel tethering in protostome and deuterostome synapses, complex and dynamic processes are likely at play. In accordance, pre-synaptic scaffolding proteins Mint1 and CASK, have been also been found to contribute to proximal coupling of Ca_v2 channels with the exocytotic machinery in both rodents (Maximov and Bezprozvanny, 2002) and the mollusc snail Lymnaea stagnalis (Spafford et al., 2003). Furthermore, unique specializations appear in distinct lineages, such as vertebrate synaptic protein interaction (“synprint”) sites in the II-III linkers of vertebrate Ca_v2.1 and Ca_v2.2 channels, which directly interact with vesicular SNARE complex proteins syntaxin-1A/B and SNAP-25 to regulate channel pre-synaptic tethering and gating (Sheng et al., 1994; Rettig et al., 1996).

As noted above, low voltage activated Ca_v3 type channels are implicated in “low threshold exocytosis,” occurring in neuroendocrine cells (Carbone et al., 2006a,b) and neurons capable of graded synaptic transmission (Weiss et al., 2012; Weiss and Zamponi, 2013). Though less well documented than spike-dependent transmission (i.e., elicited by action potentials and Ca_v2 channels), graded transmission plays important roles in certain neurophysiological contexts in both vertebrates and invertebrates. In invertebrates, LVA calcium channels and graded synaptic transmission play major roles in the activity of intrinsically rhythmic neural circuits (i.e., central pattern generators or CPGs), such as the interneuron network that drives heart contraction in the protostome leech (Angstadt and Calabrese, 1991; Lu et al., 1997). In vertebrates, Ca_v3 channels also contribute to graded synaptic transmission, including in the retina, and between neurons located in the central and peripheral nervous systems (Weiss and Zamponi, 2013). The first detailed description of graded vs. spike-dependent transmission came from leech interneurons of the heart CPG, where classical spike-dependent synaptic transmission, driven by an HVA Ca_v channel (perhaps Ca_v2), was found to co-exist in the same neurons with graded transmission driven by an LVA Ca_v channel (perhaps Ca_v3) (Angstadt and Calabrese, 1991; Lu et al., 1997). Here, two alternate modes of cellular excitability, action potentials vs. sub-threshold plateau potentials, were respectively found to elicit strong or graded inhibitory post-synaptic responses between paired CPG neurons. Subsequently, similar bimodal cellular excitability and synaptic transmission was documented in the vertebrate olfactory bulb (Egger et al., 2003).

Interestingly, “bimodal” excitability and graded vs. spike-dependent synaptic transmission have also been documented in the neuromuscular junction of medusozoan jellyfish Aglantha digitale. Here, neuromuscular synapses of large axon motor neurons manifest either low-threshold, spike-independent synaptic transmission, which elicit graded contractions of the bell myoepithelium during slow pelagic swimming, or spike-dependent synaptic transmission, triggered by aggressive, predatory tactile cues, which elicit strong contractions of the bell during fast escape swimming (Mackie, 1980; Roberts and Mackie, 1980; Kerfoot et al., 1985; Mackie and Meech, 1985; Meech and Mackie, 1993). During pelagic swimming, spontaneous depolarizing synaptic inputs into motor neurons activate an LVA channel resembling a Ca_v3 type, generating low threshold Ca²⁺ spikes with peak depolarization just below action potential threshold (i.e., about −25 millivolts or mV). These subthreshold Ca²⁺ spikes occur as spontaneous bursts at a rate of about 3–4 per second (Meech and Mackie, 1993), and trigger exocytosis and mild graded contractions in myoepithelial striated muscle cells. In the same axons, strong depolarizing sensory inputs generate Na⁺-dependent action potentials, which trigger all-or-none exocytosis and transmission producing much stronger contraction of the bell myoepithelium for the escape response. Of note, the data does not rule out the possibility that the Aglantha Ca_v3-like channel activates an HVA Ca_v channel which in turn associates with the exocytotic machinery. However, the peak of the LVA Ca²⁺ spike only reaches about −25 mV (Meech and Mackie, 1993), which is barely at the activation threshold for recorded HVA Ca_v channels from cnidarians (Przysiezniak and Spencer, 1992; Jeziorski et al., 1998). Thus, the LVA calcium channel observed in Aglantha axons might well be positioned within nanometer proximity of the exocytotic machinery, able to directly activate exocytosis. Such an association is not without precedent: The three vertebrate Ca_v3 channel isotypes were recently found to directly interact with core SNARE proteins syntaxin-1A (all three channel isotypes) and SNAP-25 (only Ca_v3.2), and disruption of the syntaxin-Ca_v3.2 channel interaction was found to abrogate the channel's contribution to low-threshold exocytosis in a neuroendocrine cell line (Weiss et al., 2012).

Whereas both Ca_v1 and Ca_v2 type channels are expressed in bilaterian neurons and neuroendocrine cells, Ca_v1 channels are often the only type found in smooth, cardiac, and striated muscle, with a few instances of Ca_v3 channel expression (Ren et al., 1998; Jeziorski et al., 2000; Jospin et al., 2002; Catterall, 2011; Senatore et al., 2014). Thus, in most smooth and cardiac muscle cells, Ca²⁺ influx through L-type/Ca_v1 channels serves to directly activate contractile proteins, and to trigger further increases in cytoplasmic Ca²⁺ by activating ryanodine receptors in the sarco/endoplasmic reticulum (SER) (i.e., calcium-induced calcium release or CICR)(Reuter, 1979; Tsien, 1983; Bers, 2002). In vertebrate skeletal muscle, Ca_v1 channels have evolved a specialized ability to sidestep the CICR process. Here, membrane-localized Ca_v1 channels directly interact with SER ryanodine receptors; activation of Ca_v1 channels at the membrane relays conformational changes in the ryanodine receptor leading to release of SER Ca²⁺, without a need for Ca_v1 channel Ca²⁺ influx (Tanabe et al., 1990, 1993; Catterall, 1991).

Interestingly, the coupling of cytoplasmic Ca²⁺ influx with rapid activation of contractile proteins seems to be a metazoan innovation. A recent in-depth genomic study found that the Ca²⁺- calmodulin (CaM)—myosin light chain kinase (MLCK) cascade, critical for excitation-contraction coupling, occurs strictly in metazoans where MLCK is absent in the genomes of choanoflagellates and other non-metazoan organisms (Steinmetz et al., 2012). Also interesting is that although “core” contractile proteins appear to have been present prior to the emergence of Metazoa, key proteins associated specifically with striated muscle in bilaterians are absent in cnidarians and ctenophores, which also possess striated muscle. As such, fast-twitching striated muscle likely evolved independently between at least bilaterians and cnidarians/ctenophores (Burton, 2008; Steinmetz et al., 2012).

Outside of a few species of swimming sea anemones, anthozoans are mostly devoid of striated muscle for contractile movement, bearing primitive smooth muscle cells with roles in feeding (mouth and tentacles) and digestion/reproduction (gastrovascular cavity) (Chapman, 1974; Burton, 2008). Medusozoans, which can become motile medusae (i.e., jellyfish), possess extensive striated musculature (“muscle sheets”) in the bell epithelium for swimming, in addition to smooth muscle cells (Chapman, 1974; Burton, 2008). The striated swimming muscles seem to lack extensive SER structures (Chapman, 1974; Keough and Summers, 1976; Singla, 1978; Spencer, 1979), suggesting a reduced dependency on CICR in lieu of Ca²⁺ influx through plasma membrane Ca²⁺ channels. Accordingly, removal of external Ca²⁺ during intracellular recording of bell striated muscle cells from Aglantha completely abrogates muscle action potentials and contraction (Kerfoot et al., 1985). Notably, the action potentials of these muscle cells are slow to reach peak (Kerfoot et al., 1985) compared to the Na⁺-dependent action potentials of their pre-synaptic effector neurons (Mackie and Meech, 1985; Meech and Mackie, 1993). This suggests the absence of fast, Na⁺-selective Na_v channels for depolarization (i.e., Na_v2.5-like channels, Barzilai et al., 2012). Instead, the long-lasting and complex waveform of Aglantha muscle action potentials (Roberts and Mackie, 1980) indicates that multiple Ca²⁺ conductances are at play, perhaps for endowing the cells with a capacity to respond to bimodal pre-synaptic innervation as discussed above (Kerfoot et al., 1985).

Long-lasting action potentials have also been observed in striated swimming muscle of another hydrozoan medusa, P. penicillatus. Here, intracellular recordings revealed action potentials whose depolarization depends on both Na⁺ and Ca²⁺ influx (Spencer and Satterlie, 1981). Like in Aglantha swim muscles, these spikes also exhibits long lasting plateau phases, reminiscent of vertebrate cardiac muscle in which prolonged Ca²⁺ influx through Ca_v1 channels ensures effective contraction of the heart for expulsion of blood (Grant, 2009). Given the similarity in waveforms between vertebrate cardiac muscle and jellyfish swimming muscle, and the predominance of Ca_v1 channels in driving bilaterian muscle contraction, it is tempting to speculate that the single Ca_v1 channel in cnidarians also drives muscle contraction. However, the data is sparse in this regard and other types of Ca²⁺ permeable channels could certainly be involved. As noted earlier, the most detailed molecular description of cnidarian Ca_v1 channels comes from the cloning and ectopic expression of a homolog from schyphozoan jellyfish C. capillata (Jeziorski et al., 1998). When expressed in Xenopus oocytes, Cyanea Ca_v1 behaves like protostome and deuterostome Ca_v1 channels by exhibiting high voltage of activation/inactivation, as well as apparent Ca²⁺-dependent inactivation evidenced by more rapid decay of its macroscopic currents in the presence of external Ca²⁺ compared to Ba²⁺.

Cnidarians get their name from cnidocytes, or “stinging cells,” best known for their role in jellyfish tentacles where they discharge thread-like tubules laced with painful and sometimes lethal toxins for defense and predation. As cnidocytes can only be used once, their discharge is highly regulated, especially those involved in prey capture (Anderson and Bouchard, 2009). Regulation involves a convergence of chemosensory and mechanosensory neural synaptic inputs (Pantin, 1942; Westfall, 2004; Anderson and Bouchard, 2009), intrinsic mechano- and chemo-sensitivity of the cnidocytes themselves (Brinkmann et al., 1996; Thurm et al., 1998, 2004), and cnidocyte-cnidocyte communication either directly via gap junctions (Mire et al., 2000; Price and Anderson, 2006), or through local synaptic circuits between cnidocytes and intermediate sensory cells located nearby (Holtmann and Thurm, 2001a). Notably, only hydrozoans and perhaps anthozoan sea anemones bear gap junction genes (Putnam et al., 2007; Chapman et al., 2010; Shinzato et al., 2011; Baumgarten et al., 2015), and electrical coupling between cnidocytes need not occur, even when gap junction genes are likely present (Holtmann and Thurm, 2001a,b).

Intracellular recording has revealed that cnidocytes are highly electrically active. Application of species-specific prey extracts via perfusion in sea water saline evokes depolarizing synaptic potentials and bursts of action potentials in impaled tentacle cnidocytes (Brinkmann et al., 1996; Price and Anderson, 2006); perfusion of Ca²⁺-free saline and calcium channel blocker Ni²⁺ disrupt this induced activity, likely through pre-synaptic disruption Ca_v channels and exocytosis (Price and Anderson, 2006). Interestingly the cnidocyst, an endomembrane-derived organelle harboring the cnidocyte stinging thread, is thought to resemble synaptic and neuroendocrine vesicles in that its exocytotic discharge depends on both membrane depolarization and Ca²⁺ influx (Skaer, 1973; Gitter et al., 1994). Patch clamp recording of cnidocytes has failed to directly identify Ca_v channel currents, but this has been attributed to washing out of endogenous currents during patch clamp recording (Anderson and Bouchard, 2009). Furthermore, in situ localization of Ca_v channel transcripts in N. vectensis revealed that one of the three Ca_v2 channel isotypes, Ca_v2a, is strongly expressed in cnidocytes (Moran and Zakon, 2014), and both a full length Ca_vβ subunit and a fragment of an unspecified Ca_v channel have been detected in cnidocyte-specific mRNA from the Portugese man'o war (Physalia physalis) (Bouchard et al., 2006; Dunn, 2009). Thus, it is likely that in vivo Ca_v channels contribute to exocytosis of the cnidocyst, perhaps using similar machinery used for pre-synaptic and neuroendocrine secretion. However, cnidocyst exocytosis likely involves additional molecular adaptations that prevent exocytosis in the absence of proper chemosensory and mechanosensory inputs (Anderson and Bouchard, 2009).

Finally, Ca_v channels have been implicated in calcification of corals, which accumulate CaCl₂ at a rate of about 10 kg per meter squared of coral reef per year (Chave et al., 1972). Application of phenylalkylamine and dihydropyridine Ca_v1 channel blockers attenuates calcification in corals Stylophora pistillata (Tanbutté et al., 1996) and Galaxea fascicularis (Marshall, 1996), and a cloned Ca_v1 channel gene was detected as an expressed protein in the calicoblastic ectoderm of Stylophora, which is involved in calcium carbonate precipitation (Zoccola et al., 1999). Thus, for at least some species of coral, Ca_v channels might contribute to the calcification process, but certainly other Ca²⁺-handling channels, pumps and exchangers also likely contribute (Marshall, 1996; Tanbutté et al., 1996).

Of the six documented Trichoplax cell types, gland cells most resemble neurons and neurosecretory cells in their expression of exocytotic SNARE proteins and membrane-apposed vesicles (Syed and Schierwater, 2002; Smith et al., 2014). Gland cells are concentrated around the periphery of the flat irregularly-shaped animal (Figure 5A), with some sparsely distributed along the ciliated ventral epithelium. Interestingly, gland cell vesicles exhibit cell-specific variability in electron density when observed under transmission electron microscopy (Smith et al., 2014). Thus, some gland cells appear specialized for secreting regulatory peptides, many of which are predicted from the genome (Nikitin, 2015) (i.e., those bearing electron dense vesicles), while others perhaps transmit small molecules such as amino acids and biogenic amines (Srivastava et al., 2008) (pale vesicles). Heterogeneity of gland cell vesicular content is also apparent through immunolabeling with antibodies against RFamide and FMRFamide neuropeptides (Schuchert, 1993; Smith et al., 2014), which selectively label a subset of gland cells located on the animal's periphery. Indeed, key questions remain about the role of gland cells in Trichoplax physiology and behavior, however their apparent exocytotic nature, coupled with their peripheral localization, suggests that they exert paracrine control over other cells for coordinating activity (Smith et al., 2014, 2015). For example, Trichoplax feeding behavior begins with the animal pausing over patches of algae detected under its body, via coordinated cessation of beating monociliated cells on the ventral epithelium (Ueda et al., 1999; Smith et al., 2014). The detected algae are then lysed by hydrolytic secretions from lipophil cells, but only those positioned close by. The animal remains sessile after algal lysis, anchored along the periphery while internally located cells undergo a churning movement. Afterwards, ventral epithelial cells resume beating, and the animal glides on to new algae (Smith et al., 2015). Here, peripherally-located gland cells are optimally positioned to exocytose paracrine factors over the entire ventral epithelium (Smith et al., 2014, 2015). Instead, more centrally located gland cells are proposed to function as chemosensory cells, which exocytose factors to inform adjacent lipophil cells of the presence of algae (Singla and Reiter, 2006; Smith et al., 2015). Further work needs to be done to understand how Trichoplax cells are coordinated in the absence of synapses during feeding and other behaviors. However, the expression of SNARE proteins in gland cells, combined with the presence of Ca²⁺-sensitive elements of the exocytotic machinery in the genome (Srivastava et al., 2008) (e.g., synaptotagmin, synaptophysin, and complexin), hints that these particular cells exhibit Ca²⁺-dependent exocytosis.

Our ongoing research on Trichoplax Ca_v channels provides indirect support for the potential dependence on calcium for exocytosis in gland cells. We recently cloned and in vitro expressed the single T-type Ca²⁺ channel homolog from Trichoplax, finding that despite more than 600 million years of divergence from vertebrate channels, it bears the distinguishing structural features of this channel type including a selectivity filter motif of EEDD, and a predicted helix-loop-helix gating brake structure in the domain I-II linker (Figure 2B). Also conserved with other T-type channels are its biophysical properties, where in vitro recorded Ca²⁺ currents exhibit hallmark attributes including a low voltage of activation, rapid activation and inactivation kinetics (Figures 5B,C), and a reduced selectivity for Ca²⁺ over Na⁺ compared to HVA Ca_v1 and Ca_v2 channels (in review). Thus, it seems as though the basic structural and functional features of Ca_v3 channels, optimized for regulating excitability and driving low threshold exocytosis, were established very early on during evolution and perhaps extend beyond Trichoplax to the even more primitive homolog identified in the genome of choanoflagellate Salpingoecca rosetta (Fairclough et al., 2013; Moran and Zakon, 2014).

Interestingly, immunoabeling with specific custom antibodies against TCa_v3 produced labeling exclusively in gland cells, with intense staining along the outside edges where vesicular exocytosis is likely to take place (Smith et al., 2014). We also recently cloned and in vitro expressed the Trichoplax Ca_v2 channel homolog, and, similarly, immunolabeling points to exclusive expression in gland cells, along with the HVA subunit Ca_vα₂δ (unpublished data). Thus, beyond the presence of exocytotic machinery and “neurosecretory” ultrastructural markers, gland cells resemble select neurons and neurosecretory cells in their expression of both HVA (Ca_v2) and LVA (Ca_v3) channels (Weiss and Zamponi, 2013). It will certainly be interesting to determine whether gland cells exhibit homologous interactions between Ca_v2 and Ca_v3 channels and proteins which in neurons and neurosecretory cells complex the channels within nanometers of the exocytotic machinery, as discussed above (e.g., RIM, Mint1 and CASK, and the SNARE proteins). We note from an ongoing transcriptomic study that mRNAs of numerous pre- and post-synaptic scaffolding genes are indeed expressed in Trichoplax, including RIM, Mint1 and CASK (Figure 5B). Also of interest will be to determine how gland cells and other cell types might be electrically activated in the absence of synaptic inputs, perhaps via paracrine signaling, or, via cell-intrinsic sensory pathways as occurs in cnidarian cnidocytes. Indeed, the nature and purpose of electrical signaling in Trichoplax remains a mystery. However, it is likely of significant importance, where in addition to Ca_v channels, the animal expresses mRNAs of a core set of genes required for generating action potentials and propagating fast electrical signals: One of the two Na_v2 channels predicted from the genome (Srivastava et al., 2008; Liebeskind et al., 2011), K_v channels of the Shaker, Shab, Shal, and Shaw varieties, a Ca²⁺-activated K⁺ channel (i.e., large conductance BK), a K_v channel accessory β subunit, and 2-pore K⁺ (K_2P) leak channels and inward rectifying K⁺ (K_IR) channels, essential for setting the polarized resting membrane potential of excitable cells (Figure 5D).

It is interesting that of the three Ca_v channel types, the Ca_v1 channel appears to be the most highly expressed in the Trichoplax transcriptome (Figure 5D), considering the specialized role that Ca_v1 channels play in excitation-contraction coupling in muscle, and the absence of ultrastructural markers for muscle in Trichoplax (Smith et al., 2014). The animal possesses the core genetic elements required for the establishment and operation of rapidly contracting muscle cells, and in fact shares slightly more of these genes with bilaterians/cnidarians than do ctenophores (Steinmetz et al., 2012), which appear to have independently evolved muscle (Ryan et al., 2013). Still debated is whether Trichoplax represents a simplified animal (Ryan and Chiodin, 2015), where the absence of clear ultrastructural markers for muscle might reflect a lost or diminished phenotype. However, the retention of contractile genes in the genome indicates utility, where they might play roles in contractile cellular processes nonetheless, or perhaps serve completely different functions. In support of the former, Trichoplax performs movements that appear independent of the ciliated ventral epithelium, such as the churning motion during feeding, and folding or rippling along its edges (Heyland et al., 2014; Smith et al., 2015). It might be the case that rudimentary contractile mechanisms underlie these movements; fiber cells, which lie between the dorsal and ventral epithelia and have branched protrusions that contact all other cell types, have been proposed to mediate contractile movements (Schierwater, 2005). However, whether these or any other Trichoplax cell types employ contractile genes in a manner homologous to muscle remains to be determined, as is the role for the single Ca_v1 channel.

Trichoplax ciliary locomotion presents distinct modalities, including starting and stopping, as well as rotation and direction changes (Ueda et al., 1999; Heyland et al., 2014; Smith et al., 2015). Although poorly understood, transitions in Trichoplax locomotive states are dependent on food concentration (Ueda et al., 1999), and might require altered beating modes of cilia that project from the ventral epithelium. In other eukaryotic cells, alterations in ciliary waveform depend on Ca_v channels (Quarmby, 2009). Well documented examples of alternate modes for ciliary beating come from ctenophores (discussed below), as well as single-celled protists such as paramecia and the green algae Chlamydomonas reinhardtii, which alter ciliary waveforms in response to external stimuli, temporarily changing swimming trajectory. Early studies on paramecia revealed that their ability to reverse upon mechanical stimulation relies on external Ca²⁺, whose transient influx specifically into cilia triggers a switch in the beat cycle (Naitoh, 1968; Kung and Naitoh, 1973). Ciliary beating in paramecia is referred to as the “ciliary” waveform, which consists of an asymmetric power stroke in one direction followed by a weaker recovery stroke in the other. The ciliary waveform is common in metazoans, exemplified in the human lung where epithelial cilia use this pattern to expel particles and fluid into the pharynx (Satir and Christensen, 2007). In paramecia, electrophysiological recording revealed that the calcium channels responsible for switching the direction of the ciliary power and recovery strokes, and hence direction of movement, reside exclusively along the cilia and not the cell body (Dunlap, 1977). Chlamydomonas also exhibits direction changes, where light or mechanical stimulation causes a switch in ciliary waveform. Normally, the two cilia of Chlamydomonas exhibit ciliary waveforms with the power strokes that pull the cell body forward; upon optical/mechanical stimulation, and in a Ca²⁺-dependent manner, the cilia switch to a symmetrical “flagellar” waveform (Bessen et al., 1980), much like that of swimming sperm, temporarily reversing movement such that cell body leads while the cilia push from behind. Here, light or mechanical stimulation generates a depolarizing membrane potential (Harz and Hegemann, 1991), activating a pre-metazoan Ca_v channel homolog positioned along the distal portion of the cilia (Fujiu et al., 2009), consistent with the calcium channel localization in paramecia cilia. Gene disruption of the Chlamydomonas channel, dubbed CAV2, abrogates both the light and mechanical induced reversal, indicating a convergence of the two sensory modalities on CAV2 channel activation (Matsuda et al., 1998; Fujiu et al., 2009).

The mechanisms by which Ca²⁺ influx controls ciliary waveform transition involves dynamic regulation of dynein motor complexes positioned between pairs of ciliary microtubules (Yang et al., 2001; Hayashi et al., 2002; Patel-King et al., 2002; Wargo and Smith, 2003, 2007). Notably, Ca_v channels and Ca²⁺ influx are not necessary for maintaining ciliary beating per se, and might be specific for altering waveforms under transient, induced conditions (Tamm, 1994, 2014a). A recent study looking at the roles of Ca_v channels in ciliary beating of mammalian ependymal cells, which move cerebral spinal fluid in the central nervous system, found no effect of Ca²⁺ influx on ciliary beating and fluid movement, where Ca_v1 channels were found localized mostly in the cell soma (Doerner et al., 2015). Instead, Ca²⁺ influx through CatSper channels in sperm triggers hyperactivation of the flagellar beat (Qi et al., 2007). Indeed, whether the cilia on Trichoplax's dorsal and ventral epithelium exhibit alternate or modulated beating modes remains to be determined; if so, it will be interesting to evaluate whether Ca_v channels are involved.

Placozoans and cellular sponges both exhibit quasi-coordinated contractile behavior in the complete absence of neurons and muscle. However, whereas Trichoplax has the majority of genes required for excitation-contraction coupling, the absence of Na_v and K_v channels in sponges precludes rapid fluctuations in membrane potential, at least by canonical means. Sponges do possess K_2P leak (Wells et al., 2012) and inward rectifying (Tompkins-Macdonald et al., 2009) K⁺ channels, which if expressed in contractile cells would establish negative resting membrane potentials. However, in fresh and sea water sponges from the genus Microciona, increasing external [K⁺], which diminishes the K⁺ membrane gradient and would depolarize cells, has no bearing on contraction (Prosser, 1967). However, contraction does depend on the presence of external cations, which presumably move across the cell membrane, albeit in a non-selective manner: External Na⁺ can be substituted with K⁺ or Li⁺, and Ca²⁺ can be replaced with Mg²⁺ or Sr²⁺. Notable is that cellular contractility appears highly atypical in this clade, in its dependence on both external Ca²⁺ and Mg²⁺, but perhaps more Mg²⁺, being roughly 5-fold more concentrated than Ca²⁺ in sea water. Instead, specimens from the genus Euspongia do show a selective dependency on external Na⁺ and Ca²⁺, and increasing external [K⁺] triggers marked and prolonged contraction consistent with depolarization-induced contractility (Pavans de Ceccatty, 1971). Nonetheless, these contraction events occur in the absence of measurable electrical impulses, and their slow and long-lasting kinetics make it unlikely that voltage-gated channels play a role.

The absence of gap junctions in sponges, combined with the absence of electrical signaling in cellular sponge species in general, indicate that contraction and its propagation from cell-to-cell occurs though much slower cellular pathways. Some have therefore speculated that contractile waves spread along sponge tissues by means of paracrine secretion, where incoming paracrine factors cause cells to both contract and to secrete (Leys and Meech, 2006). With respect to cellular contraction, its dependency on extracellular cations implies that ions move across the membrane, perhaps through ion channels, pumps and/or exchangers which are regulated by ligand-dependent receptors such as G-protein coupled receptors (GPCRs). Several “slow” GPCR pathways exist in muscle which dynamically regulate myosin light chain phosphorylation/dephosphorylation and hence contraction of actin/myosin filaments (Somlyo and Somlyo, 2003). In addition, some GPCR pathways directly regulate muscle contraction and tone through release of Ca²⁺ from internal stores and/or activation of other muscle effector kinases such as protein kinase C and Rho-associated protein kinase (Sanderson et al., 2008); in this context, membrane Ca²⁺ influx is thought necessary only for the replenishment of internal stores, and not for regulating contraction per se. Thus, contractile activity in the absence of fast electrical signaling is not unprecedented. Beyond pinacocytes and actinocytes, other sponge cells exhibit extensive motility, which depends on influx of extracellular calcium and could contribute to gross body movement via cumulative action of multiple cells (Lorenz et al., 1996).

Glass sponges (class Hexactinellida) are the only poriferan group known to exhibit electrical signaling, in the form of Ca²⁺ action potentials which travel through a multinucleated syncytium comprising the entire body (Leys and Mackie, 1997; Leys and Meech, 2006). Although some cells remain separate, they are nevertheless connected to the syncytium though cytoplasmic bridges (Mackie, 1981), making the entire glass sponge body one large electrically conductive system. Unlike cellular sponges, glass sponges are incapable of contracting, and as noted above, prevent particulates from entering their aquiferous systems by arresting choanocyte ciliary beating and the feeding current (Mackie, 1979; Lawn et al., 1981). The entire process, from stimulus onset to current cessation, occurs within roughly 20 s and can be triggered either spontaneously, in response to excessive sediment in the water, or via applied mechanical or electrical stimulation (Lawn et al., 1981; Mackie et al., 1983). External recording of stimulated tissue reveals a biphasic action potential, with a depolarizing inward cation current preceding a repolarizing current presumably carried by efflux of K⁺ ions (Leys and Mackie, 1997; Leys et al., 1999). Voltage-gated Ca²⁺ channels are thought to mediate action potential depolarization, since reduction of external [Na⁺] to 25% of physiological levels only minimally affects the action potential, whereas application of Ca_v channel blockers Co²⁺ and Mn²⁺ and Nimodipine significantly disrupt it (Leys et al., 1999). Given the loss of Na_v2 channels in sponges, the single Ca_v channel found in the sponge genome is a viable molecular candidate for driving the glass sponge action potential. Not clear is how repolarization takes place in the absence of K_v channels. Application of K_v channel blocker tetraethylammonium (TEA) delays and diminishes the amplitude of the action potential, but it does not specifically prolong the depolarization phase, which would be expected if blocking a repolarizing K_v channel current. Given the slow kinetics of the action potential, it might be that repolarization involves slower K⁺ conductances, such as K_2P or K_IR channels. Furthermore, contributions to repolarization might be attributed to accumulated inactivation of the voltage-gated Ca²⁺ channel, consistent with the observed refractory period of the sponge action potential of roughly 29 s (Leys et al., 1999).

Also not clear is how propagating Ca²⁺ action potentials lead to arrest of choanocyte ciliary beating. Based on the observed involvement of calcium channels in regulating Chlamydomonas and protozoan ciliary beating, Ca_v channels and Ca²⁺ influx have been suggested to play a role (Leys and Meech, 2006), however, their localization to choanocytes, and the Ca²⁺-sensitivity of choanocyte-driven feeding current, have yet to be explored. Likewise, involvement of the sponge Ca_v channel in altered ciliary beating of larvae in response to light stimulation has not been explored, where perhaps the mechanisms underlying both choanocyte ciliary arrest and larval ciliary switch use overlapping mechanisms.

Phylogenetic analyses suggest that ctenophores are the most basal animals (Dunn et al., 2008; Ryan et al., 2013; Moroz et al., 2014; Whelan et al., 2015), leading to the controversial hypothesis that they independently evolved a nervous system (Marlow and Arendt, 2014; Moroz et al., 2014; Moroz and Kohn, 2015, 2016; Ryan and Chiodin, 2015). The most identifiable feature of the nervous system are synapses, defined by the presence of pre-synaptic active zones, connected across the synaptic cleft to electron-dense post-synaptic densities (Heuser and Reese, 1977). Genome sequencing has revealed that the vast majority of genes involved in synapse formation and function are present in ctenophores (Ryan et al., 2013; Moroz et al., 2014; Moroz and Kohn, 2015), animals which lack synapses [sponges (Srivastava et al., 2010) and placozoans (Srivastava et al., 2008)], and even unicellular organisms that pre-date animals (King et al., 2008; Fairclough et al., 2013). Thus, the mere presence or absence of select synaptic genes, in particular those with more generalized functions in exocytosis not specific to synapses, is not enough to confirm or refute the independent evolution hypothesis, since most of these genes were present before the ctenophore divergence. Resolving this issue will require extensive molecular comparative analyses of nervous system development and function.

The ctenophore nervous system consists of two distinct nerve nets (Jager et al., 2011), as well as separate neural structures that innervate peripheral structures such as the tentacles, comb rows and mouth (Hernandez-Nicaise, 1973b, 1991; Mackie et al., 1992; Tamm and Tamm, 1995; Jager et al., 2011). Giant axons have been documented running under the comb rows of select species, which synapse onto comb plate polster cells and are believed to alter ciliary beating during escape swimming (Mackie et al., 1992). Here, the large-diameter axons are considered an adaptation to increase action potential velocity, with speeds greater than 0.5 ms⁻¹ comparable to the 1.4 ms⁻¹ of giant axons of the jellyfish A. digitale escape system (Mackie and Meech, 1985). Large diameter axons have also been documented in the mouths of beroids, which synapse onto smooth muscle and adhesive epithelial cells to coordinate swallowing of prey and the subsequent tight closure of the mouth (Tamm and Tamm, 1995). Thus, fast neuronal signaling is certainly confirmed in ctenophores, but the mechanisms by which these signals are transduced across the synaptic cleft remains a mystery. Indeed, the atypical ultrastructure of the ctenophore synapse suggests that the underlying mechanisms for synaptic transmission might be inherently different from other animals. Most notable is that in ctenophores pre-synaptic vesicles appear to bud off from the SER (or perhaps fuse with it), suggesting that vesicles are derived independently of the Golgi network and pre-synaptic endosomes, as in bilaterian synapses (Heuser and Reese, 1973; Jahn and Fasshauer, 2012). If this proves to be true, it would represent a striking fundamental difference in synapse organization and function consistent with independent evolution. By extension, this arrangement would provide ER-derived neuropeptides direct access to pre-synaptic vesicles, consistent with the notion that ctenophore synapses rely heavily on neuropeptides as transmitters (Jager et al., 2011; Moroz et al., 2014; Moroz and Kohn, 2016).

The presence of both SER and large mitochondria within nanometer proximity of the putative vesicle release sites suggests that Ca²⁺ plays a role in ctenophore synaptic exocytosis, where the two organellar systems are poised to act as sources or sinks for Ca²⁺ ions (Clapham, 2007). Unfortunately, only a few examples of electrophysiologically-recorded synaptic potentials are available for ctenophores (Moss and Tamm, 1987; Meech, 2015), and no experimental evidence is available describing the involvement of the single Ca_v2-like channel or other Ca²⁺ channels in vesicle exocytosis. In bilaterian synapses, Ca²⁺ influx through Ca_v2 channels is required to activate Ca²⁺-sensitive exocytotic machinery, and this is achieved by close apposition of the channels with docked vesicles. As noted above, some interactions with scaffolding proteins that help tether Ca_v2 channels at the synapse appear to have deep ancestry, occurring in both protostome and deuterostome bilaterians (e.g., Rim-1, Mint1, and CASK). If these interactions extend to both cnidarians and ctenophores, this would strengthen the argument for the single origin of the nervous system hypothesis. Instead, the obligate close apposition of SER and mitochondria to what appear to be docked vesicles in the ctenophore synapse might indicate that Ca_v channel nanodomain coupling is circumvented in lieu of calcium-induced calcium release from the SER, a mechanism for synaptic transmission that would be highly atypical. Clearly, this area of research warrants further study, to explain homology or convergence between synapses in ctenophores and those in cnidarians/bilaterians.

The most detailed electrophysiological records of membrane ion currents in ctenophores come from enzymatically-isolated smooth muscle cells. These cells are multinucleated and can be massive (upwards of 6 cm long and 40 μm in diameter), and are innervated by nerve net neurons at neuromuscular junctions bearing the presynaptic triad arrangement (Hernandez-Nicaise, 1973a). Isolated muscle cells from the cydippid Pleurobrachia bachei conduct action potentials and contract in response to external perfusion of L-glutamate, as less so L-aspartate, but not to other transmitters such as GABA, histamine and acetylcholine (Moroz et al., 2014). Thus, ctenophore neuromuscular junctions are proposed to be glutamatergic, although the various ionotropic glutamate receptors present in the genome have yet to be localized to muscle synapses. Intracellular voltage clamp experiments reveal diverse pools of ion channels in ctenophore smooth muscle, which generate distinct action potential waveforms in different cell types. In Mnemiopsis giant muscle cells, which project from the statocyst to the mouth and auricles (Anderson, 1984; Hernandez-Nicaise et al., 1984), the depolarization phase of the action potential is driven by at least two distinct voltage-gated channels permeable to Ca²⁺ and Na²⁺ (but more selective for Ca²⁺), both high voltage activated and sensitive to Ca_v channel blockers Cd²⁺, Co²⁺ and dihydropyridines verapamil and methoxyverapamil (Anderson, 1984; Dubas et al., 1988). One notable distinction was their kinetics, with one channel bearing fast inactivation producing fast, transient Ca²⁺ currents, and the other much slower inactivation, producing slow, long-lasting currents (Dubas et al., 1988). Similar depolarizing Ca²⁺/Na⁺ currents were reported for the action potential of giant muscle cells isolated from B. ovata (Hernandez-Nicaise et al., 1980, 1982; Bilbaut et al., 1988a,b). For both species, Ca²⁺ influx through the channels seems required for contraction, since removal of external Ca²⁺ or pharmacological disruption abrogates muscle action potentials and contractions (Hernandez-Nicaise et al., 1980; Anderson, 1984; Bilbaut et al., 1988a,b; Dubas et al., 1988; Cario et al., 1996). Unfortunately, while the available data provides convincing evidence for the existence of distinct voltage-gated Ca²⁺ channels present in ctenophore smooth muscle, little can be said about the specific channel types at play. The three most likely candidates, the single Ca_v2 and two Na_v2 channels, are expected to be highly divergent from their homologs in other animals, and their specific pharmacological, ion selectivity and biophysical properties are completely unknown. Further confounding the matter is that currents recorded from a cloned Na_v2 channel (from honeybee) are sensitive to Cd²⁺ but not dihydropyridines (Gosselin-Badaroudine et al., 2016), while in situ currents recorded from cnidarian neurons are sensitive to both Cd²⁺ and dihydropyridines (Anderson, 1987; Spafford et al., 1996). Ca²⁺ vs. Na⁺ selectivity is also not a good marker for channel identity, since it can be quite variable, a fact made evident by the T-type channel from snail L. stagnalis which becomes highly Na⁺ permeable via alternative splicing in the domain II P-loop region, without altering its Ca²⁺ selectivity filter motif of EEDD (Senatore et al., 2014). Nonetheless, it is worth noting that ctenophores are the only animals with bona fide muscle cells lacking Ca_v1 channels, the main drivers for excitation-contraction coupling in other animals. In this regard, and in accordance with the proposed independent evolution of ctenophore muscle (Ryan et al., 2013; Moroz et al., 2014), smooth muscle cells seem to depend on somewhat atypical depolarizing conductances for excitation-contraction coupling.

In B. ovata, movements of the mouth and pharynx during swallowing of prey are encoded by different conductance/contractile profiles of distinct muscle cell types, in lieu of a complex nervous system capable of sophisticated temporal and spatial synaptic outputs (Bilbaut et al., 1988a,b). Radial smooth muscle cells, which span the mesoglea and are anchored in the outer epidermis (ectoderm) and pharyngeal endoderm (Hernandez-Nicaise et al., 1980), exhibit narrow, transient action potentials required in bursts for contraction to occur. Instead, longitudinal muscles running along the ectoderm exhibit longer lasting action potentials, each capable of causing contraction. The differences in action potential waveforms are attributed to different repolarizing conductances: Radial fibers bear pronounced, rapidly activating K⁺ currents, while longitudinal fibers bear a slow Ca²⁺-activated K⁺ current (K_Ca), and a transient, voltage-sensitive K⁺ current. Worth noting is that the functional coupling of voltage-gated Ca²⁺ channels with Ca²⁺-sensitive K⁺ channels (e.g., K_Ca channels such as BK and SK), also documented in Mnemiopsis smooth muscle (Anderson, 1984), predates animals, observed in protists (Valentine et al., 2012) and dinoflagellates (Pozdnyakov and Skarlato, 2015). In vertebrates, both BK and SK K_Ca channels physically and functionally couple with Ca_v1, Ca_v2, and Ca_v3 channels, presumably to overcome the limited diffusion range of Ca²⁺ ions in the cytoplasm (Clapham, 2007; Guéguinou et al., 2014). This coupling is observed in neurons which undergo spike frequency adaptation, where trains of action potentials accumulate more and more cytoplasmic Ca²⁺ and K_Ca channel activation, leading to a slowing down of action potential frequency and an eventual disruption of the action potential spike train (Yarom et al., 1985). In accordance, longitudinal but not radial muscle in Beroe exhibit spike frequency adaptation and eventual cessation of induced spikes. As noted by the authors, the difference in muscle properties would permit bursts of synaptic inputs from the nerve net to cause short lived contractions of longitudinal fibers, while causing facilitating, long lasting contractions of radial fibers, permitting more complex movement of the mouth and pharynx (Bilbaut et al., 1988a).

It is unknown whether membrane Ca²⁺ influx alone activates contractile myofilaments, or whether internal stores from the sarcoplasmic reticulum or mitochondria contribute via CICR or some other mechanism. Notably, the sarcoplasmic reticulum (SR) is diminished compared to other smooth muscle cells, making up less than 1% of the total cell volume, and that the plasma membrane (sarcolemma) lacks typical invaginations and appositions with the SR (Hernandez-Nicaise and Amsellem, 1980; Hernandez-Nicaise et al., 1980, 1984), both of which serve to enhance the CICR process. Furthermore, in larger cells, the sarcolemma can be as far as 2000 nm away from the SR and mitochondria(Hernandez-Nicaise et al., 1980). Considering the observed range of Ca_v channel cytoplasmic Ca²⁺ plumes of roughly 100 nm, even in elevated external [Ca²⁺] (Weber et al., 2010), this separation seems rather imposing. Nevertheless, ATP-hydrolyzing enzymes that would shuffle Ca²⁺ ions into the SR and mitochondria, as well as out of the cell through the sarcolemma, have been detected (Cario et al., 1996), and Ca²⁺ fluorescence experiments reveal that action potentials trigger Ca²⁺ release from internal stores (Cario et al., 1995a,b). Thus, CICR likely takes place in ctenophore smooth muscle at least to some degree. Interestingly, all smooth muscle cells observed possess extracellular axonemes (i.e., bare cilia), which run along the lengths of the muscle fibers nestled within circumferential invaginations of the sarcolemma (Tamm and Tamm, 1989). The mechanism by which these structures are formed, and their function, remains a mystery. Given the propensity of cilia to bind Ca²⁺, they have been proposed to act as external Ca²⁺ sources/sinks for excitation-contraction coupling (Tamm, 2014a).

Ctenophores also possess striated muscle. Cydippid specimens of the genus Euplokamis possess tentacles with specialized repeating side branches (tentilla) which extend and retract to lure and capture prey with their colloblasts (Chun, 1880). Tentilla are heavily innervated, and exhibit complex localized sensory integration, where tactile or electrical stimulation can activate rapid extension of single tentilla through contraction of bundled, long striated muscle cells running along their lengths; subsequent retraction and coiling is thought to occur passively, though the elastic properties of the underlying tissue (Mackie et al., 1988). Activation of this muscle is clearly through excitation, since depolarization of the membrane with high external [K⁺] or electrical stimulation causes contraction (Mackie et al., 1988). In vivo, excitation-contraction coupling is likely elicited at neuromuscular junctions located along the outside of the bundled fibers, with few mitochondria indicating low energy expenditure as expected given the infrequent use of the tentilla for prey capture. Repeating striations in the muscle are clearly evident, bearing distinct Z, I, and A bands characteristic of striated muscle, but they lack H bands. Ctenophores lack key genes associated with striated muscle formation and function, including those involved in Z-disc formation (Steinmetz et al., 2012). Thus, striated muscle in Euplokamis might represent a third case of independently evolved striated muscle, along with bilateria and cnidaria. Like giant smooth muscle cells of Beroe and Mnemiopsis, Euplokamis striated muscle cells lack extensive SR at their center, as well as any obvious transverse tubular systems. However, they do possess a second arrangement of SR located within a few hundreds on nanometers from the sarcolemma. The involvement of membrane Ca²⁺ influx through voltage-gated calcium channels and CICR in contraction has not yet been examined, however the presence of synapses, the rapid speed of contraction, and electrical activation of these striated muscle cells imply these processes do occur.

On their own, comb plates are static and require external inputs to initiate beating (Sleigh, 1974; Tamm, 1980). A major source of activation arrives from tracks of ciliated epidermal cells, called ciliated grooves, which transduce gravitational signals from the statocyst to the first comb plate of each row (Chun, 1880) (Figure 6A). The four balancers of the statocyst each consist of compound cilia, which at their tips support a conglomerated mass of living cells called the statolith, much like four legs supporting a table. Angular body displacement causes the statolith to exert differential gravitational force on each of the balancers, altering their respective beating frequency according to the angle of force exertion (Tamm, 1982). The balancers act as pacemakers for the comb rows via mechanical coordination (Tamm, 1982) (Figure 6A). During negative phototactic swimming, when ctenophores seek swim to the surface mouth facing up, tilting toward the horizontal causes balancers (and hence comb rows) below the midline to beat more frequently than ones above, while the opposite occurs during downward, mouth down (i.e., positive geotactic) swimming. Once the animals are vertical, all eight comb rows beat at a similar frequency (Tamm, 1980, 1982). Interestingly, evidence suggests that deflection-induced changes in balancer beating requires membrane excitation and Ca²⁺ influx of through voltage-gated channels, since removal of external Ca²⁺, or application of non-specific calcium channel blockers Co²⁺ and Ni²⁺, disrupt deflection-induced responses of the balancers (Lowe, 1997). Furthermore, chemical depolarization of isolated balancers (via increasing external [K⁺]) directly increases beating frequency independent of mechanical stimulation, but only in the presence of external Ca²⁺ specifically at the base of the cilium. Thus, a proposed model for statolith activation of balancer beating is that deflections activate cationic stretch-receptors at the cell membrane, which in turn activate voltage-gated calcium channels at the base of the cilium. Ca²⁺ influx through these channels then activates ciliary beating. Notably, such an arrangement appears inconsistent with sperm flagella, where hyperactivation (i.e., an alteration of ciliary waveform) depends on the voltage-gated Ca²⁺ channel CatSper localized along the length of the cilium, not at the base. Noted above, in Chlamydononas, a Ca_v channel homolog dubbed CAV2 causes flagellar waveform change, and is also localized strictly to the distal regions of the cilia. A notable distinction between ctenophore balancers and these two other systems is that in balancers, Ca²⁺ influx at the base serves to activate ciliary beating, whereas in sperm and Chlamydomonas, distal Ca²⁺ influx along the ciliary membrane serves to alter the waveform of ciliary beating, either by increasing asymmetry of the flagellar waveform (sperm), or by increasing waveform symmetry (Chlamydomonas). Consistent with this distinction, activation of distal voltage-gated calcium channels in comb plate cilia leads to reversal of beating during backward swimming and feeding behavior, whereas voltage-gated Ca²⁺ channels located at the base of beroid macrocilia activate beating. These similarities prompted Tamm to propose conserved mechanisms for voltage-gated Ca²⁺ channel regulation of ciliary beating, where channels located along the length of the cilia influence “reprogramming” responses of the ciliary waveform (e.g., reversal, waveform changes), whereas those at the base of the cilia influence “on-off” responses (e.g., activation, arrest, or increase in beating frequency) (Tamm, 1994, 2014a).

The propensity of ctenophores to switch between upward swimming (negative geotaxis) and downward swimming (positive geotaxis) is referred to as their “mood,” and is thought to be regulated by neural inputs to balancer cells from various sensory modalities, including those tuned to water disturbances and hydrostatic pressure (Tamm, 1982, 2014a; Lowe, 1997). Electron microscopy of the statocyst and surrounding areas reveals neurites which synapse onto balancer cells(Tamm, 1982; Hernandez-Nicaise, 1991). Furthermore, intact larvae and dissected statocysts from M. leidyi and Pleurobrachia pileus respond to ectopic electrical stimulation and membrane depolarization with high external [K⁺] by switching between geotactic states (Lowe, 1997), consistent with regulation by electrical signaling. Mechanistically, a switch in mood/geotactic state requires that the same deflectional forces acting on the balancer cilia produce opposite effects on their beating frequency at different times. Nonetheless, regardless of geotactic sign, ciliary deflection in the appropriate direction causes Ca²⁺ influx and increased beating. Thus, the mechanisms by which putative synaptic inputs alter geotactic mood in the balancers are likely independent of the stretch-activated channels and voltage-gated Ca²⁺ channels. Finally, ctenophores are able to override geotactic behavior and exhibit different types of swimming, such as horizontal, feeding, and reverse escape. The mechanisms by which this takes place are not known (Tamm, 2014a); neural inputs to the balancers, ciliated grooves, and/or comb plates is a possiblility.

In Euplokamis, electrical stimulation near the mouth causes a temporary reversal of comb plate beating and thus reverse swimming, presumably via nerve impulses from giant axons running under the comb plates which synapse onto polster cells (Mackie et al., 1992). Instead, stimulation at the aboral end causes increased ciliary beating for fast-forward swimming, which based on experiments in Pleurobrachia, might occur via nerve inputs downstream of the statocyst perhaps at the ciliated grooves or the aboral-most comb plates (Tamm, 1982). Similar fast-forward and reverse swimming responses are observed for Mnemiopsis upon contact with a jellyfish predator (Kreps et al., 1997). Unilateral reversal of comb plate beating is also observed during cydippid feeding (Tamm and Moss, 1985), where in Pleurobrachia, comb rows flanking tentacles with ensnared prey reverse, leading to rotation of the animal such that the appropriate tentacle bearing food approaches the mouth. Here, direct electrical stimulation of a single tentacle or its adjacent body surface causes comb plate reversal, through an apparently bilateral conduction pathway innervating only the four ipsilateral comb rows of that tentacle (Moss and Tamm, 1993). The mechanism for comb plate reversal is thought to occur directly in comb plate polster cells, where intracellular recording at the cell soma revealed that neural stimulus-induced synaptic potentials give rise Ca²⁺-dependent action potentials, leading to ciliary reversal (Moss and Tamm, 1986). Extracellular recording and Ca²⁺ imaging of comb plate cilia revealed that action potentials propagate from base to tip of the cilium (Tamm and Terasaki, 1994), and could be abrogated by application of inorganic calcium channel blockers (Moss and Tamm, 1987). Thus, similar to Chlamydomonas CAV2, a voltage-gated Ca²⁺ channel distributed along the ciliary membrane mediates motor responses. It will be interesting if this channel turns out to be Ca_v2, suggesting an ancient and conserved coupling of Ca_v channels types with ciliary reversal.

Finally, giant “toothed” macrocilia located inside the mouth of predatory beroid species receive synaptic input from a giant axon nerve net (Tamm and Tamm, 1985, 1995), whose excitatory inputs activate beating from an otherwise intermittent/quiescent state during engulfing of prey (Tamm, 1983, 1988a). The macrociliary power stroke is directed into the body cavity, helping to draw or macerate prey into the stomach (Swanberg, 1974). Notably, both semi-intact preparations of macrocilia, as well as isolated macrociliary cells, become activated in response to depolarization with high external [K⁺], but only in the presence of external Ca²⁺ which need only be applied by perfusion to the base of the cilium (Tamm, 1988a,b). Furthermore, application of non-selective calcium channel blockers prevents macrociliary activation (Tamm, 1988a). Indeed, all of the data is consistent with a model where synaptic inputs depolarize the membrane to activate voltage-gated Ca²⁺ channels strictly at the base of the cilia to initiate beating (Tamm, 2014a). Thus, either differential localization of the same voltage-gated calcium channel used to reverse beating of comb plate cilia, or a different channel altogether (perhaps also used at the base of balancers), plays the role of activating beating of macrocilia, consistent with Tamm's hypothesis on localization-dependent “on-off” vs. “reprogramming” function of ciliary calcium channels.