Mechanisms and physiological implications of cooperative gating of clustered ion channels

Abstract

Ion channels play a central role in the regulation of nearly every cellular process. Dating back to the classic 1952 Hodgkin–Huxley model of the generation of the action potential, ion channels have always been thought of as independent agents. A myriad of recent experimental findings exploiting advances in electrophysiology, structural biology, and imaging techniques, however, have posed a serious challenge to this long-held axiom, as several classes of ion channels appear to open and close in a coordinated, cooperative manner. Ion channel cooperativity ranges from variable-sized oligomeric cooperative gating in voltage-gated, dihydropyridine-sensitive Ca_V1.2 and Ca_V1.3 channels to obligatory dimeric assembly and gating of voltage-gated Na_V1.5 channels. Potassium channels, transient receptor potential channels, hyperpolarization cyclic nucleotide-activated channels, ryanodine receptors (RyRs), and inositol trisphosphate receptors (IP₃Rs) have also been shown to gate cooperatively. The implications of cooperative gating of these ion channels range from fine-tuning excitation-contraction coupling in muscle cells to regulating cardiac function and vascular tone, to modulation of action potential and conduction velocity in neurons and cardiac cells, and to control of pacemaking activity in the heart. In this review, we discuss the mechanisms leading to cooperative gating of ion channels, their physiological consequences, and how alterations in cooperative gating of ion channels may induce a range of clinically significant pathologies.

CLINICAL HIGHLIGHTS

• Ion channels regulate multiple physiological processes including excitability, contraction, neurotransmitter release, and gene expression. Changes in the function of many channels have been linked to pathological conditions such as arrhythmias, epilepsy, and hypertension.

• In this review, we discuss a vast and growing body of work that suggests that many ion channels do not gate independently, as originally assumed. Instead, many Ca²⁺, Na⁺, and K⁺ channels seem to open and close cooperatively.

• The physiological and pathological implications of cooperative gating are profound. For example, positive cooperativity amplifies Na⁺ and Ca²⁺ entry into cardiac muscle. Enhanced cooperativity between clustered dihydropyridine-sensitive Ca²⁺ channels has been suggested to be critical for sympathetic stimulation of the heart, but when excessive it causes cardiac arrhythmias and hypertension.

• Enhanced cooperativity among K_Na1.1 channels has been linked to epilepsy.

• The development of rational strategies for the targeted modulation of ion channel cooperativity may allow tuning of Ca²⁺ and electrical signaling in health and disease.

1. INTRODUCTION

Nearly 70 years ago, Hodgkin and Huxley established the foundations of modern biophysics with the advent of their quantitative model for action potential (AP) generation and propagation in neurons (1). This seminal work has been subsequently modified and expanded to form the basis of our understanding of electrical signaling in all excitable cells (2). A central tenet of the Hodgkin–Huxley model was that individual channels gate independently and that the macrocurrents recorded across the cell membrane reflect the ensemble behavior of these quantal conductance elements. Direct testing of the model in general, and the independence principle specifically, became feasible with the advent of the patch-clamp technique (3, 4), which permitted investigators to obtain ultra-low noise current records with high temporal resolution from a small area of membrane (<1 μm²) produced by the putative opening of single ion channels.

Although the Hodgkin–Huxley model accommodates the random simultaneous openings of multiple channels, the probability of a set of channels behaving synchronously is expected to decrease as the number of channels in the set increases. A simple way to explain this is to use the example of flipping a “fair” coin, in which each flip is independent from the other. The probability of observing a tail or a head is 0.5 for every trial, but the probability of observing two tails in a row is 0.5² (i.e., 0.25), that of three tails in a row is 0.5³ (i.e., 0.13), and so on. If individual membrane channels behave in an analogous independent fashion, then the current amplitude frequency distributions will be fit with a Poisson function (5). Similarly, the timing between successive channel openings should be uncorrelated. Many ion channels act this way, further fueling the idea that ion channels gate randomly and independently from one another (6).

A limitation of the patch-clamp technique is that it monitors a membrane area of <1 μm², severely limiting the number of channels that are sampled. Because most ion channels are randomly distributed in the surface membrane, the probability of a patch pipette encompassing a region of membrane containing a group of homogeneous channels is relatively low. Accordingly, recordings of the concerted openings of multiple channels are generally rare but were reported as early as the 1990s for Na⁺, hyperpolarization-activated cyclic nucleotide-gated (HCN), and Ca²⁺ channels (7–13).

The development of electron-multiplying charged-coupled device (EMCCD) cameras has facilitated the recording of optical Ca²⁺ signals from single channels or small clusters of channels with ultrahigh temporal and spatial resolution (14–17). This has afforded direct, detailed observation of rare and low-amplitude Ca²⁺ signals resulting from the openings of Ca²⁺-permeable ion channels in the surface membrane and intracellular organelles from relatively large areas of the cell (compared with patch clamp). Indeed, imaging systems such as confocal and evanescent field total internal reflection fluorescence (TIRF) microscopes have been used by multiple groups to detect Ca²⁺ influx events via small clusters of Ca²⁺-permeable channels in a wide range of cells (14, 16, 18–26), revealing that cooperative gating of Ca²⁺ channels is not as rare as previously thought.

At this point, it is important to distinguish the terms “cooperative gating” and “functional coupling” of ion channels. Cooperative gating of ion channels refers to when two or more ion channels from the group open in concert because of physical interactions with one another within a cluster. In this gating modality, gating of one channel is allosterically communicated to other attached channels, increasing their probability of gating. Functional coupling of ion channels refers to interactions between different ion channels in macromolecular complexes displaying cross talk with one another in signaling microdomains. A primary example of functional coupling is the action of Ca²⁺ flux through one channel group that acts as a second messenger to facilitate gating of a molecularly distinct group of ion channels.

The principal focus of this review is to present the available evidence for cooperative gating of ion channels. We focus on key classes of membrane channels for which there is substantial evidence for this gating modality. We discuss putative mechanisms by which the cooperativity is produced, the physiological implications of cooperative gating in excitable tissues, and possible pathologies that may result from alterations in this behavior. Finally, we identify key gaps in our present knowledge and propose new research avenues for this emerging field of inquiry.

2. STOCHASTIC SELF-ASSEMBLY OF ION CHANNEL CLUSTERS

All models of cooperative gating of ion channels proposed to date are predicated on having the same ion channel aggregated into dense clusters where the centroids of adjacent members are separated by slightly more than the radius of their cross-sectional areas (5, 27, 28). This proximity between the same ion channels within clusters allows physical interactions between adjacent channels to occur, leading to amplification of channel function. Thus, it is pertinent to discuss current views on how clusters of membrane channels are generated.

The complex process by which ion channels are inserted into a membrane begins with the transcription of the genes encoding the pore-forming and accessory subunits into messenger RNA (mRNA). These mRNAs are subsequently translated by ribosomes and transfer RNA (tRNA) molecules associated with the endoplasmic reticulum (ER) to polypeptides. The growing polypeptides of transmembrane channel subunits are progressively translocated and folded into the ER membrane. Once the full protein is synthesized and integrated in the membrane, a vesicle containing this protein forms and fuses with the cis Golgi face (FIGURE 1). As the channel proteins transit through the Golgi stack, they eventually reach the trans Golgi network (TGN). Through this process they undergo posttranslational processing, including N-linked glycosylation. Ion channels leave the Golgi apparatus via vesicles that are transported by molecular motors moving on microtubules that cross over the trans Golgi network and are anchored on the plasma membrane. These vesicles eventually fuse with the surface membrane (29, 30). Once they are delivered to the surface membrane, gating of the new channels can regulate excitability as well as multiple signaling cascades. The ion channels will remain in the plasma membrane until they are removed and either recycled or degraded via endocytic pathways (31, 32).

FIGURE 1.

Ion channel proteins translocate to the membrane of the rough endoplasmic reticulum (ER), traffic to the Golgi, and are transported to the surface membrane on microtubules, where they form clusters. Left: cartoon depicting the synthesis, membrane translocation, processing, and trafficking of ion channels to the surface membrane. Top right: superresolution image of ion channel clusters in the surface membrane. Bottom right: histogram of ion channel cluster areas from images like that above. The purple, green, and gray ovals at the tip of the microtubules cartoon represent microtubule-anchoring/binding proteins such as BIN1. Histogram was modified from Sato et al. (5) with permission. Cartoon created with Biorender.com.

There is a large and still growing body of evidence showing that the distributions of ion channels vary along a cell’s membranous surface (33–36). The studies that have examined how ion channels organize in the plasma membrane involved detailed analyses of electron micrographs (35–39), TIRF and confocal images (40), and, more recently, superresolution microscopy (12, 28, 41–46). The majority of ion channels examined to date aggregate into dense clusters (see FIGURE 1). For example, in neurons, N-methyl-d-aspartate (NMDA) channels aggregate in dendritic and somatic membranes (47). Similarly, voltage-gated Na⁺, Ca²⁺, and K⁺ channels that produce action potentials required for neurotransmitter release are specifically expressed in the axon terminals (33, 34, 48, 49). The same is true for striated and smooth muscle, where dihydropyridine-sensitive voltage-gated Ca²⁺ channels form clusters within the sarcolemma of these cells (36, 38, 42, 45, 46, 50–52).

This clustering of ion channels seems to be critical for multiple physiological processes in both neurons and muscle tissue. Indeed, the clustering of voltage-gated Ca²⁺ channels is necessary for the amplification of Ca²⁺ influx that is required to trigger the release neurotransmitter at neuronal terminals and to initiate the process of excitation-contraction (EC) coupling in muscle (11, 15, 24). Furthermore, recent studies indicate that ion channels involved in cooperative signaling cascades cocluster. A notable example is the coclustering of large-conductance Ca²⁺-activated K⁺ (BK) channels and Ca_V1.3 channels in neurons (41, 53–55). In this case, cooperative gating of Ca_V1.3 channels clusters provides the necessary Ca²⁺ signal to activate nearby BK channel clusters. We discuss additional examples of this type of organization and signaling modality later in this review (see sects. 4.4 and 8.3).

Recently, Sato et al. (5) and Baddeley et al. (28) implemented a combination of in silico and experimental approaches to investigate the mechanisms regulating channel clustering in the membrane. They found that the distributions of clusters of six types of channels [i.e., Ca_V1.2, BK, transient receptor potential (TRP) vanilloid (TRPV)4, Ca_V1.3_S, Ca_V1.3_L, and ryanodine receptors (RyRs)] were all described by a single exponential function regardless of their expression in different cell types (FIGURE 1). Inositol 1,4,5-trisphosphate receptors (IP₃Rs) have similar distributions in the sarcoplasmic reticulum (SR) and endoplasmic reticulum (56–58). These studies suggest that the presence of Ca_V1.2, BK, TRPV4, Ca_V1.3_S, RyR, and IP₃R clusters in cellular membranes may be the result of a stochastic self-assembly process. Whether active mechanisms of aggregation of ion channels may also contribute to this process is currently unclear. Moreover, although it is possible that not all channels are organized into clusters via stochastic self-assembly processes (59–63), the preponderance of the data suggests that clustering is the default mode of organization of ion channels in the plasma membrane as well as in the endo/sarcoplasmic reticulum of cells. In sects. 3–10, we describe current understanding of clustering and cooperative gating in several classes of ion channels.

3. VOLTAGE-GATED Ca²⁺ CHANNELS

Voltage-gated Ca²⁺ (Ca_V) channels are expressed in excitable cells and nonexcitable cells (reviewed in Ref. 64), where they respond to electrical depolarizations by shifting their voltage sensors upward to open the channel. This movement and/or the resultant Ca²⁺ influx through the channel pore is an essential trigger for many physiological processes including contraction of skeletal, smooth, and cardiac muscle, hormone secretion, regulation of gene expression, and neurotransmission. The structure and function of Ca_V channels have been extensively reviewed elsewhere (65–74) and are thus only briefly addressed here.

Ca_V channels, like the evolutionarily related voltage-gated Na⁺ (Na_V) channels, contain a large, pore-forming α₁-subunit that distinguishes the channel within this 10-member channel family (FIGURE 2A) (69). The 10 types of Ca_V channels are further categorized into three subfamilies, Ca_V1, Ca_V2, and Ca_V3, based on their pharmacological and biophysical profiles as well as their amino acid sequence similarity (75). Channels within each subfamily share >70% homology, whereas across subfamilies the channels are <40% homologous (75). The α₁-subunits are 170- to 250-kDa (65, 73) single polypeptides, with intracellularly located NH₂ and COOH termini flanking four repeating domains (DI–DIV), each with six transmembrane segments (S1–S6). Voltage sensitivity is conferred by segments S1–S4 (termed the voltage sensing domain; VSD), and S5–S6 and the associated loop from each of the four repeats combine to form the pore domain. Recent cryo-electron microscopy (cryo-EM) studies of purified skeletal muscle Ca_V1.1 channels have revealed that the four repeat domains of α_1S are arranged in a clockwise manner about the central pore (76, 77). This spatial arrangement is thought to be conserved throughout the eukaryotic Ca_V and Na_V channel families (72).

FIGURE 2.

Structure and dimensions of voltage-gated Ca²⁺ (Ca_V) channels. A: illustration of the pore-forming Ca_Vα₁ and auxiliary Ca_Vβ, Ca_Vα₂δ, and Ca_Vγ. AID, alpha-interacting domain. B: 2 illustrations of patch pipettes forming a gigaseal enabling voltage control and single-channel recordings from the underlying patch of membrane. The pipette may enclose either a circle (i) or a 3-dimensional Ω-shaped patch of membrane (ii) that may contain a large number of ion channels. C: dimensions of Ca_V1.1 (PDB: 5GJV) from Wu et al. (76) (left) and the topological view/footprint of a single α_1S pore-forming subunit (right), illustrating that many channels can theoretically be housed in a single patch. D: the dimensions of a single Ca_V1.1 tetrad (left) and an array of 4 tetrads (right). Figure created with Biorender.com.

Although the α₁-subunits dictate the main channel properties, Ca_V1 and Ca_V2 channels are multimeric complexes comprised of the α₁-subunit along with up to three auxiliary (or accessory) subunits that can modify channel kinetics and play roles in channel trafficking and regulation (FIGURE 2A) (78, 79). These subunits include the cytosolic Ca_Vβ, the Ca_Vα₂δ on the extracellular leaflet of the sarcolemma bilayer, and the transmembrane Ca_Vγ. Ca_V3 channels can be regulated by these accessory subunits (80–82) but do not appear to require coassembly with them (68, 83, 84).

3.1. L-Type Ca²⁺ Channels

The Ca_V1 subfamily (Ca_V1.1–Ca_V1.4) constitutes the so-called L-type calcium channels, with L signifying that they generate large, long-lasting (slowly inactivating) currents. Ca_V1 channels are expressed in myocytes, cardiac pacemaker cells, nerves, endocrine cells, the inner ear, and the retina and are characterized by their pharmacological sensitivity to organic Ca²⁺ channel blockers including dihydropyridines, phenylalkylamines, and benzothiazepines (85). They contain α₁-subunits encoded by cacna1s, c, d, and f. Functional channels are expressed in complex with 1:1 stochiometric combinations of various isoforms of Ca_Vβ, Ca_Vα₂δ, and sometimes Ca_Vγ (68–70).

Cooperative gating has been observed and reported in a diverse variety of voltage- and ligand-gated ion channels, as we comprehensively review here. It is notable, however, that in recent years major advances in our understanding of this phenomenon have been extracted from studies of voltage-gated calcium channels, and most specifically in three of the L-type calcium channels, namely Ca_V1.1, Ca_V1.2, and Ca_V1.3. We summarize the current thinking in sects. 3.2–3.7.

3.2. Ca_V1.1 Channel Structure and Function

The skeletal muscle isoform of the L-type Ca²⁺ channel was the first of the voltage-gated calcium channels to be cloned (86) and coincidentally was also the first to be proposed to exhibit cooperative gating behavior (7). These channels, still sometimes referred as 1,4-dihydropyridine receptors (DHPRs) because of their responsivity to pharmacological agonists and antagonists from this class of drugs, are abundantly expressed in skeletal muscle, where they are physically tethered to juxtapositioned type 1 ryanodine receptors (RyR1s) in the triad junctions. Freeze-fracture electron microscopy studies revealed that Ca_V1.1s are arranged into tetrads (4 channel groups), with each individual Ca_V1.1 aligning with a single RyR1 homotetramer (87). The RyR1s are arranged in a checkerboard formation, where every other RyR1 channel is associated with a Ca_V1.1 tetrad and in which the RyR1s are also reported to exhibit cooperative gating behavior (88) (see sect. 7.2). Because of the physical coupling between the t-tubule-localized Ca_V1.1 and the sarcoplasmic reticulum (SR)-localized RyR1, Ca²⁺ influx through Ca_V1.1 is not mandatory for excitation-contraction (EC) coupling (89). Instead, the action potential-driven movement of the Ca_V1.1 voltage sensors is allosterically conveyed to the attached RyR1, inducing a conformational change in the RyR1 that results in Ca²⁺ release from the SR and subsequent activation of the myofilaments. This sarcolemmal Ca²⁺ influx-independent form of RyR1-mediated Ca²⁺ entry, where Ca_V1.1 serves solely as a voltage sensor, has been termed excitation-coupled Ca²⁺ entry (ECCE) (90). The details of this physical interaction and mechanical gating mechanism are still not fully elucidated despite decades of work to reveal to essential interaction sites. However, recent near-atomic-resolution structural studies of both Ca_V1.1 (76, 77) and RyR1 (91) have led to the development of a speculative theory on how this may occur (92). Although the details remain unresolved, it is definitively known that there are at least four critical protein components of the triad, including Ca_V1.1, RyR1, auxiliary subunit of calcium channels Ca_Vβ_1a [the one and only Ca_Vβ isoform expressed in skeletal muscle (93)], and adaptor protein STAC3. The absence of any of one of these proteins precludes EC coupling (reviewed in Ref. 94). Pathophysiologies associated with mutations in Ca_V1.1 include hypokalemic periodic paralysis (HypoPP-1) and malignant hyperthermia (MHS5) (68).

3.3. Cooperativity Among Ca_V1.1 Channels

Αn early paper in this field from the Schindler group reported cooperativity and clustering of purified Ca_V1.1 channels when they were isolated from guinea pig skeletal muscle and reconstituted into planar lipid bilayers (7). They observed a time dependence of channel conductance in which larger channel conductance levels emerged over time after bilayer formation: Immediately after planar bilayer formation, channel openings with a 0.9-pS conductance were observed (with 100 mM BaCl₂ as the charge carrier in the pipette). This result was suspected to represent the monomeric channel conductance. However, as time passed after membrane formation and more channels were incorporated that presumably had time to cluster and functionally interact, larger conductance values emerged corresponding to oligomeric channel openings of 4, 8, 16, 24, 32, 40, up to 64 channels. Given the propensity of these channels to form tetragonal arrays, it is interesting that each of these oligomeric openings is an even-integer quadruplet of the ∼1-pS single-channel conductance. Independently gating channel openings are predicted to follow a binomial distribution such that, in a multichannel patch, one would expect to observe most frequently a conductance equivalent to a single channel opening, with multiple channel openings occurring with progressively lower frequencies. However, binomial analysis of the conductance amplitude frequency histogram from Ca_V1.1 channels revealed a positive deviation, indicative of cooperative gating, where the 40-pS conductance was observed to occur more frequently than 4-, 24-, or 32-pS conductance levels. This preference for the 40-pS conductance, equivalent to a simultaneous oligomeric opening of 10 channels, is highly statistically improbable if these channels are independently gating entities.

At first glance, the idea that 64 channels might be captured in a cell-attached patch may seem unlikely. However, some simple calculations confirm that this is not such an unlikely occurrence if one assumes a pipette tip diameter of 1 µm and an area of membrane under the patch that should initially be well approximated by the area of a circle (πr²). However, upon gigaseal as a three-dimensional Ω-shaped section of membrane is pulled into the pipette, the area may be better described by a sphere (4πr²) (FIGURE 2B). Thus, the patch area could be 0.79–3.14 µm². A key question then is, how many channels or channel tetrads could fit into that space? For single channels, cryo-EM structures of Ca_V1.1 have revealed their dimensions, confirming their widest aspect as 100 Å (10 nm) (76, 77) (FIGURE 2C). Assuming then that a channel takes up 10 × 10 nm² = 100 nm² = 0.0001 µm² and that they are as densely packed as possible, 7,900–34,400 channels could theoretically occupy the area of a typical patch. Even assuming 1/100th of that density, that still allows 64 channels to easily fit into the area of a typical patch. Knowledge accumulated from EM studies suggests that a Ca_V1.1 tetrad can fit into a 30 × 30 nm² = 300 nm² = 0.0003 µm² area (95–97) (FIGURE 2D). We also know that tetrads line up with every other RyR1. If we assume a tetragonal array of four tetrads, the dimensions would be 90 × 90 nm² = 8,100 nm² = 0.0081 µm² (FIGURE 2D). To get to 64 channels, that would be 16 tetrads with an assumed array dimension of 120 × 210 nm² = 25,200 nm² = 0.0252 µm², which could still easily fit into the area of a typical patch.

Unfortunately, this paper from the Schindler group (7) is still largely ignored by the skeletal muscle and Ca_V1.1 field, but it has gained traction in recent years, as evidenced by an increasing number of citations as other groups began to report cooperativity in a diverse variety of ion channels. However, Ca_V1.1 cooperative gating has not been revisited. We speculate that this may be explained by the fact that, historically, Ca_V1.1 channels have been difficult to express in heterologous expression systems. This difficulty has recently been obviated by the finding that coexpression of STAC3 permits robust trafficking and recording of these currents in mammalian cell lines including tsA-201 (98). With the emergence of this accessible heterologous expression system that permits the study of Ca_V1.1 channels in near isolation, the path is clear for a fresh study on Ca_V1.1 cooperativity. Indeed, many interesting questions as to how the channels interact remain to be answered, including 1) Is the interaction a direct physical coupling between adjacent channels such that the conformational change that occurs when one channel opens is allosterically conveyed to the attached channels, increasing their open probability (P_o), or is it an indirect association that requires the presence of another protein that could bridge adjacent channels together? 2) How do adjacent tetrads interact? and 3) Is channel cooperativity affected by receptor signaling cascades? These questions should be addressed in future studies of Ca_V1.1. Addressing these questions will pave the way to determining whether cooperative gating of native Ca_V1.1 channels regulates skeletal muscle EC coupling.

3.4. Ca_V1.2 Channel Structure and Function

Ca_V1.2 was the second of the L-type channels to be cloned (99), hence the “2” in its name. This chronological nomenclature is maintained through the remainder of the L-type channel subfamily and indeed for each of the three voltage-gated calcium channel subfamilies (75). The α_1C-subunit constitutes the pore-forming core of these L-type Ca²⁺ channels that forms multimers with at least Ca_Vα₂δ and Ca_Vβ [with Ca_Vβ₂ being the most prevalent isoform in the heart, in particular Ca_Vβ_2b (100)] and Ca_Vβ₃ in smooth muscle (101) (FIGURE 2A). Associations of Ca_Vα_1C with Ca_Vγ are still unclear in primary cells but do appear to occur in transiently transfected HEK293 cells (102). Despite the Ca_V1.2 channels often being referred to as the cardiac L-type Ca²⁺ channel, perhaps because they were originally cloned by screening cardiac cDNA libraries for homology with the skeletal muscle isoform, this is somewhat of an exclusionary misnomer given that Ca_V1.2 channels are expressed in a variety of tissues including vascular and visceral smooth muscle, brain, pancreas, adrenal gland, and, of course, cardiac muscle (65, 68). Within these tissues, Ca_V1.2 channels are involved in a wide range of fundamental physiological processes including cardiac (103–105) and smooth muscle (15, 74, 104, 106–109) EC coupling, insulin secretion via excitation-secretion coupling (104, 110, 111), and regulation of gene expression via excitation-transcription (ET) coupling (24, 112–115). In the brain, they shape neuronal firing and support neuronal plasticity, learning, and memory formation (116, 117). The critical role of Ca_V1.2 in many of these processes has been confirmed in mouse models with tissue-specific Ca_V1.2 knockout (110, 118–120) or in mice with a single T1066Y mutation in their Ca_Vα_1C that renders the channels dihydropyridine insensitive (104). This second strategy has been especially helpful in tissues where both Ca_V1.2 and Ca_V1.3 are expressed, as there remains no established pharmacological means to unequivocally distinguish between these two channels.

Ca_Vα_1C is extensively alternatively spliced, generating several splice variants that create even more structural and functional diversity in these channels (as reviewed in Refs. 121–123). Smooth muscle Ca_Vα_1C is predominantly expressed with exon 8, whereas the most highly expressed cardiac Ca_Vα_1C has exon 8a instead. Exon 8 confers higher sensitivity to dihydropyridine calcium channel-blocking drugs (123), contributing to their efficacy as antihypertensives. Exon 1 is also spliced to form at least three variants, a long NH₂-terminal-encoding 1a (expressed in cardiomyocytes), a short NH₂-terminal 1b (expressed in smooth muscle and brain), and an even shorter 1c (expressed in cerebral arteries) (65). The alternative splicing of exon 1 is also thought to alter sensitivity to the nondihydropyridine calcium channel blocker diltiazem. Accordingly, the smooth muscle variant 1b is about twice as sensitive to diltiazem than the cardiac 1a (123).

In cardiac muscle, influx of Ca²⁺ through Ca_V1.2 channels is obligatory, as the channels are not mechanically linked to the type 2 cardiac ryanodine receptors (RyR2s) that cluster on the other side of the 12- to 15-nm dyadic cleft (124, 125) on the junctional sarcoplasmic reticulum (jSR). Accordingly, instead of acting exclusively as a voltage sensor, cardiac Ca_V1.2 channels must open in response to membrane depolarization as the action potential propagates through the myocardium. This allows a small amount of Ca²⁺ to enter the cell. Influx of Ca²⁺ through a single Ca_V1.2 channel is known as a “Ca²⁺ sparklet” and can be imaged with Ca²⁺-sensitive fluorescent indicators (14, 126). This Ca²⁺ entry triggers further “Ca²⁺-induced Ca²⁺ release” (CICR) from ryanodine receptors on the SR, producing a “Ca²⁺ spark” (127–132). Synchronized activation of multiple sparks across the tens of thousands of dyads in a single cardiomyocyte (133), leads to a global elevation in intracellular Ca²⁺ concentration ([Ca²⁺]_i) that activates the myofilaments and elicits contraction. Cardiac EC coupling has been extensively reviewed elsewhere (see Refs. 103, 134).

Immunogold labeling (38) and freeze-fracture electron microscopy studies (124, 135, 136) have revealed clustering of Ca_V1.2 channels in heart muscle. This clustering is now frequently examined in superresolution light microscopy studies (42, 50, 51, 137, 138). Channel clustering is a fundamental requirement for cooperative interactions between ion channels. Within clusters, channels are densely packed in close proximity of one another, making functional, physical interactions between adjacent channels plausible. Clustering of Ca_V1.2 channels in dyadic regions is thought to maximize the probability of triggering release from adjacent RyR2s during EC coupling (11, 42, 139, 140). The relationship between local Ca²⁺ influx and Ca²⁺ spark activation has been the subject of intense investigation. The majority of data support the view that, at negative potentials (−50 to −30 mV), where the driving force for Ca²⁺ influx is high, the opening of a single Ca_V1.2 channel produces a unitary current (i_Ca) of 0.5–0.4 pA (126, 141) that is sufficient to activate a spark (142–145). However, in humans, EC coupling takes place during phases 1 and 2 of the AP, when cardiomyocytes are much more depolarized (∼+50 mV). At these potentials, with supraphysiological 20 mM extracellular Ca²⁺ concentration ([Ca²⁺]_o), a single Ca_V1.2 channel opening, with an i_Ca ≈ 150 fA, can activate a spark (126). However, in physiological 2 mM Ca²⁺, the calculated i_Ca is only 10 fA (139), ∼15-fold smaller than in 20 mM Ca²⁺. Is this a sufficient Ca²⁺ influx to reliably trigger a Ca²⁺ spark and permit contraction? Work from several laboratories suggests that it is not. Inoue and Bridge (139) reported that Ca²⁺ influx occurring at the peak of the AP (+50 mV) triggered Ca²⁺ sparks with a probability (P_spark) of almost unity (0.997) in rabbit ventricular myocytes. Given the low open probability (P_o) of Ca_V1.2 channels at these potentials, they proposed that a cluster of 5–10 channels must be available to activate a spark with a high probability. Sobie and Ramay (140) and Polakova et al. (146) further concluded that because of the low i_Ca and brevity of the channel openings at these potentials, multiple channels must open simultaneously to reliably trigger a spark. However, the maximum P_o of Ca_V1.2 channels with physiological [Ca²⁺]_o is not unity (108, 139, 141, 147), with some of the higher estimates in unstimulated cells (i.e., cells examined without β-adrenergic agonists) being P_o = 0.3. If all Ca_V1.2 channels gate independently, the probability of 5–10 channels opening simultaneously is much lower than P_spark (i.e., 0.3⁵–0.3¹⁰). So, how can one reconcile this high probability of a spark with a relatively low probability that multiple Ca_V1.2 channels open at once? Cooperative gating, as we discuss in sect. 3.5, can cause clusters of cardiac Ca_V1.2 channels to open simultaneously, driven by the channel in the cluster with the largest P_o (11). Thus, cooperativity can explain this discrepancy, and we discuss the evidence for and the mechanisms underlying Ca_V1.2 channel cooperativity in sect. 3.5.

EC coupling in smooth muscle is different from that of cardiac and skeletal muscle, but one of the common elements is the essential presence and function of Ca_V1.2 channels (107, 149). Electro-mechanical coupling describes the process through which depolarization-activated Ca²⁺ influx leads to smooth muscle contraction. Ca_V1.2 channels are the primary mediators of this response, but Ca²⁺ influx across the smooth muscle sarcolemma can come from a variety of sources including T-type calcium channels, TRP channels, and Orai (74, 150–152). The incoming Ca²⁺ binds to calmodulin (CaM) with a stoichiometry of 4:1, and the resultant Ca²⁺·CaM activates myosin light chain (MLC) kinase (MLCK) that phosphorylates MLC in an ATP-dependent manner. This cascade results in actin-myosin cross-bridge formation and smooth muscle contraction (153). Demonstrating the importance of Ca_V1.2 channels for development and maintenance of vascular tone and reactivity, L-type calcium channel blockers are effective antihypertensive drugs (154).

Clustering of Ca_V1.2 channels has also been extensively reported in vascular smooth muscle, where it is also thought to facilitate interactions between adjacent channels (45, 46, 52, 155–157). Furthermore, Ca_V1.2 clustering has been reported in pancreatic β-cells at sites of secretory granule exocytosis (158), in sensory neurons from the superior cervical ganglia (159), in hippocampal neurons (49, 160–162), and in heterologous cell lines (42, 163–165). The evidence for cooperative interaction of Ca_V1.2, the mechanisms underlying it, and the physiological implications are discussed in sect. 3.5.

3.5. Cooperativity of Ca_V1.2 Channels in Health and Disease

The vast majority of the currently published data on Ca_V1.2 channel cooperativity has been collected in our laboratories, and certainly that is where the idea was first proposed and pursued. We tell the story of it here chronologically as it transpired. The first indication that Ca_V1.2 channels do not always gate in an independent manner came in 2005 when L. F. Santana’s group began using optical techniques to record Ca_V1.2 single-channel activity in vascular smooth muscle cells isolated from rat cerebral arteries (FIGURE 3A) (26). Their methods (14) entailed loading cells with a fluorescent Ca²⁺ indicator dye (in this first case it was fluo-5F) and EGTA via a patch pipette during whole cell voltage clamp, while the cells are held at −70 to −90 mV. TIRF imaging of the cell footprint is then performed at 90–100 Hz with an EMCCD camera. At the chosen membrane potentials, Ca_V1.2 channels exhibit low P_o but high driving force, such that the opening of a channel, when it does occur, generates sufficient Ca²⁺ influx (with 20 mM external Ca²⁺) to produce a resolvable signal termed a Ca²⁺ sparklet when it binds to fluo-5F. The EGTA in the dialyzed solution then buffers the Ca²⁺ and acts as a sink, preventing the generation of a global Ca²⁺ transient, which would obscure the detection of single-channel events. This powerful approach, now utilized in multiple laboratories to record sparklets from diverse Ca²⁺-conducting ion channels (12, 23, 166–170), permits visualization of single-channel activity over the entire coverslip-adhered portion of the sarcolemma, providing a much larger sampling area than one can obtain with a patch pipette recording from a cell-attached patch (17, 20). The change in Ca²⁺ over time at active regions of interest (ROIs) is then plotted to generate traces that resemble single-channel electrophysiology recordings, albeit with a reduced sampling frequency (FIGURE 3A). With this broader view in which channel activity could be observed and compared across the dorsal sarcolemma, it was obvious that not all Ca_V1.2 channels behaved in the same way. In ROIs where multiple channels were present, some unusual and unexpected gating behavior was observed in which large, time-correlated multichannel openings occurred frequently. The striking synchrony with which these groups of multiple channels opened and closed suggested that the gating of one channel was influencing the gating of the others. In these regions, the channels also appeared to have a higher P_o that generated “persistent” Ca²⁺ entry, while other sites displayed activity more consistent with the low P_o of an independently gating single channel (13, 171). Ca²⁺ sparklets were eliminated by nifedipine, and the L-type calcium channel agonist BayK 8644 increased the probability of observing a Ca²⁺ sparklet. The frequency with which openings and/or closings of multiple channels were observed in perfect unison suggested that a subset of Ca_V1.2 channels were not gating independently but rather cooperatively, or, as we called it at the time, in a “seemingly coupled” manner. We later favored the terminology “cooperative gating” over “coupled gating,” as coupled implies 2 channels, but the simultaneous openings and closings we observed were not limited to 2 channels and often involved larger groups of 6–10 channels (42). Of course, the imaging experiments had a technical limitation in that we could only sample as fast as the camera could read off the data, and even with small imaging areas our technical limit was ∼100 Hz. So, the possibility existed that we were simply sampling too slowly and missing the stepwise notches one might expect to see if a few independently gating channels opened or closed one after the other. Would we still observe these apparently simultaneous multichannel gating events in 10- to 20-kHz single-channel electrophysiology experiments? To address this issue, we made cell-attached patch-clamp electrophysiological recordings and obtained analogous results (FIGURE 3B), and thus the notion that Ca_V1.2 channels exhibited cooperative gating behavior was born (13). This contention that Ca_V1.2 channels could gate in a nonindependent manner was, and still is, somewhat controversial, given that it confutes the long-standing Hodgkin–Huxley hypothesis that ion channels gate in a mutually independent manner (1). It has certainly made for some lively discussions at the Biophysical Society Annual Meetings!

FIGURE 3.

Recording and analysis of Ca_V1.2 cooperative gating events: total internal reflection fluorescence (TIRF) microscopy configuration (A) and cell-attached patch-clamp electrophysiology (ephys) system (B) for recording Ca²⁺ sparklets and single-channel currents, respectively. [Ca²⁺]_i, intracellular Ca²⁺ concentration. Traces and histograms used to detect and analyze Ca_V1.2 cooperative openings in different cell types are modified from Navedo et al. (13) with permission. Figure created with Biorender.com.

In subsequent studies, the cooperativity hypothesis has been consistently supported and we have delved more deeply into the physiological and pathophysiological mechanisms underlying this noncanonical channel gating behavior. Notably, PKC agonists enhanced the persistent, cooperative gating activity at discrete sites in vascular smooth muscle cells but not all sites, suggesting that some channels were within the range of PKC signaling whereas others were seemingly out of range. Later work (13, 172, 173) revealed that this was because a subset of Ca_V1.2 channels associate with the A-kinase anchoring protein AKAP5 (previously known as AKAP150 in mice and as AKAP79 in humans), which anchors PKA, PKC, and calcineurin (CaN) (174, 175) next to Ca_V1.2 channels in vascular smooth muscle, cardiac muscle, and neurons (13, 45, 172, 176–178). In smooth muscle, PKCα-mediated phosphorylation elevates Ca_V1.2 activity (13, 173, 179, 180). PKCα activity is increased during angiotensin II (ANG II)-induced hypertension, and indeed both acute and chronic ANG II stimulation of smooth muscle cells isolated from mouse mesenteric and cerebral arteries lead to increased Ca²⁺ influx and increased vascular tone in intact arteries, which is at least partially driven by enhanced cooperative gating behavior of Ca_V1.2 channels (13, 173, 180). AKAP5 anchoring of PKC is essential for these effects as AKAP5−/− mice are hypotensive, do not develop hypertension in response to chronic ANG II infusion, and do not exhibit enhanced Ca²⁺ sparklet activity or elevated levels of cooperative gating (13, 173). Furthermore, pharmacological PKC activators enhance cooperative gating behavior, but this response is absent in cells isolated from AKAP5−/− arteries. These studies suggest that cooperative gating of Ca_V1.2 channels causes I_Ca amplification during angiotensin-induced hypertension.

In 2010, after many years of observing this unusual gating behavior without a well-defined mechanism to explain it, the hypothesis that Ca_V1.2 channels can gate cooperatively was finally tested in cardiac and arterial smooth muscle myocytes, and the first mechanistic details began to emerge (13). Cooperativity was visualized in Ca²⁺ sparklet and single-channel electrophysiology recordings (FIGURE 3). There are several ways to identify positive cooperativity, including binomial and variance analyses, conditional dwell time density analysis, and maximum-likelihood fitting analysis (reviewed in Ref. 181). In binomial analysis, the idea is that if channels gate in an entirely independent manner, then the probability of observing 1 channel opening should be higher than that of observing the simultaneous opening of 2 channels, and in turn there should be a progressively lower probability of observing instances of 3, 4, or 5 channels opening together… and so on. Thus, if one were to plot the all-points amplitude histogram of i_Ca, independently gating channels should generate a binomial distribution. However, with positive cooperativity, multichannel openings may occur more frequently, and thus the amplitude histogram would deviate from the binomial distribution. This indeed was the case for Ca_V1.2 channels. However, there appears to be no favored or set number of channels that preferentially undergo cooperative gating events, and often up to 10 channels can be observed to gate coordinately (13, 42). This is in stark contrast to findings in cardiac Na_V1.5 channels, where dimers of those channels seem to preferentially gate together in a 14-3-3-dependent manner (10) as discussed in sect. 5. The absence of a single preferential oligomeric state led us to pursue another metric to quantify the degree of cooperativity, namely a coupled Markov chain model. A similar approach had previously been implemented by Chung and Kennedy (182) in their analysis of partially coupled GABA-activated chloride channels. Implementation entails calculating a dimensionless coupling coefficient parameter, kappa (κ), from single-channel activity traces (13). In this model, a κ value of 1 indicates channels that always gate cooperatively, a κ value of 0 indicates channels that always gate independently, and a κ value in the range 1 > κ > 0 indicates a mixture of cooperative and independent openings. This script, written in MATLAB, is freely available as source code accompanying Ref. 42.

In a channel patch or sparklet site, variable numbers of channels open and close coordinately, alongside some independent openings of single channels, and thus the fully cooperative case is seldom observed in an experimental setting with a full sweep of activity. Ca_V1.2 channel activity in the absence of agonists or other stimulation often shows κ values between ∼0.0 and 0.4, indicating a mixture of channels gating independently and others displaying nonindependent, cooperative gating, even under basal conditions (13, 50, 165). Indeed, in a 2010 study testing the cooperative hypothesis of Ca_V1.2 channels, the median κ value reported was 0.22 (13). To explore the mechanisms underlying cooperative gating, channels lacking a portion of the COOH-terminal tail between amino acids 1670 and 2171, where AKAP5 was putatively thought to bind to a leucine zipper (LZ) motif (175), were generated, and they failed to display cooperative gating behavior. Furthermore, cooperativity was also absent in AKAP5-null cells (173). Increased fluorescence resonance energy transfer (FRET) between COOH-terminal-tagged Ca_V1.2-EGFP and Ca_V1.2-tagRFP correlated with enhanced cooperativity, whereas decreased FRET correlated with more independent gating. These results suggested that physical proximity or interactions of Ca_V1.2 COOH-terminal tails are a component of the cooperative gating mechanism. Furthermore, PKCα activators were found to enhance cooperative gating, in line with previous findings (13). A subsequent study in tsA-201 cells expressing cardiac Ca_V1.2 channels found that the dynamic trafficking of these proteins in vesicular structures contributes to cluster formation and cooperative gating behavior (29). Here, disruption of microtubules or actin elements induced a profound reduction in cooperative gating. These results suggest that an intact cytoskeleton facilitates Ca_V1.2 clustering and cooperative gating. Consistent with this idea, an intact cytoskeleton was recently found to be essential for regulation of Ca_V1.2 clustering and cooperative gating in response to β-adrenergic stimulation in adult cardiomyocytes (51).

To summarize this first advance in our still-evolving mechanistic understanding of Ca_V1.2 cooperative interactions, five insights have emerged: 1) physical proximity or interactions between the COOH-terminal tails of adjacent channels is required; 2) cooperativity occurs in the subpopulation of channels that associate with AKAP5; 3) cooperativity requires an intact cytoskeleton; 4) cooperativity is facilitated by channel phosphorylation; and 5) it is regulated by CaM (FIGURE 4).

FIGURE 4.

Mechanism of L-type Ca²⁺ channel cooperative gating. A: model of the interaction between 2 Ca_V1 channels based on structures by Wu et al. (76) and Fallon et al. (189). B: schematic showing our proposed model of L-type Ca²⁺ channel cooperative gating based on the analysis of Ca_V1.2 and Ca_V1.3 channels (12, 40). Two channels are illustrated for simplicity and are drawn bridged by Ca²⁺·calmodulin (CaM) in the manner of the 2 published crystal structures of COOH-terminal fragment dimers (189, 190). Although this is speculative at present, we do know that the interactions depend on Ca²⁺·CaM (12, 42) and intact pre-IQ motifs (12, 42) and occur in AKAP scaffolded microdomains (at least for Ca_V1.2) (13, 173). For >2 channel multimers to gate coordinately, we postulate that there are at least 2 possibilities: 1) Ca²⁺·CaM may “daisy-chain” adjacent pre-IQ motifs, utilizing the 2 CaM-binding sites on that motif, or 2) one lobe of Ca²⁺·CaM may bind to pre-IQ and the other may bind to another of the established CaM binding sites on these channels (for review see Ref. 445). In our scheme, we begin at Bi, where the membrane is at its resting potential and channels are close to one another but not interacting in the closed state. Bii: with depolarization, a subset of the channels stochastically open and Ca²⁺ flows into the cell, binding to CaM. Biii: Ca²⁺·CaM facilitates the physical, functional interactions of adjacent channels that depend on binding to the pre-IQ motif. In this functionally cooperative state, the opening of 1 channel is allosterically communicated to the attached channel and they gate coordinately. Biv: Ca_V channels undergo voltage- and Ca²⁺-dependent inactivation but remain associated for a time after the Ca²⁺ signal has decayed, leaving them in a primed state as the cells repolarize. Thus, with high-frequency activation, e.g., in the heart during fight or flight, the channels can immediately transition to the cooperative open state, resulting in an immediate facilitation of Ca²⁺ influx. During low-frequency stimulation, e.g., during resting heart rate, the physical bridges between adjacent channels in a cluster are dissolved, and the resting confirmation is assumed as the cycle is resumed. V_m, membrane potential. Figure created with Biorender.com.

The contribution of cooperative gating to pathological states was observed in several studies (11, 13, 183) in which Ca_V1.2 cooperativity was notably enhanced in channels with a gain-of-function mutation [G436R in mice, G406R in humans (71)] associated with the multisystem disorder known as Timothy syndrome (TS) or long QT 8 (LQT8) (184). The enhanced levels of cooperativity of TS channels that incompletely inactivate and thus enhance Ca²⁺ entry into ventricular myocytes hinted that the cooperativity mechanism may be facilitated by Ca²⁺ itself. This was indeed shown to be the case in 2015 when electrophysiological recordings made with Ba²⁺ as the charge carrier in place of Ca²⁺ failed to display cooperativity and physical channel-channel interactions (42).

With the light-induced fusion of Flavin-binding, Kelch repeat, F box 1 (FKF1) and gigantea (GI) proteins, it was definitively shown that physical interactions between Ca_V1.2 channel COOH-terminal tails could facilitate cooperative interactions and amplify I_Ca (11). In this FKF/GI system, which controls flowering during daylight hours in Arabidopsis thaliana, FKF1 undergoes a conformational change upon illumination with blue light that allows it to bind to GI within minutes but dissociate with a half-life (t_1/2) of ∼60 h. We tagged Ca_V1.2 channels with this system to produce Ca_V1.2-GI and Ca_V1.2-FKF and expressed them together in tsA-201 cells, finding that I_Ca was enhanced after blue light (11). Furthermore, channel activity and cooperativity were enhanced after blue light. One of the clear advantages of this system is that it permits recording control currents first and then stimulating with blue light to observe the effects of physical channel-channel interactions in the same cells. Unfortunately, the dissociation of this system is too slow to allow visualization and recording of the unbinding effects, but there are now several, more recently characterized light-activated dimerization systems that dissociate on a more manageable timescale, including the cryptochrome 2 and CIBN system, which reverses with a t_1/2 of ∼5.5 min (185, 186). Expression of the GI- and FKF1-fused wild-type (WT) channels in rat ventricular myocytes with biolistic transfection (also known as “gene-gun”) led to larger electric field stimulation (EFS)-evoked Ca²⁺ transients and enhanced cell shortening after blue light-induced dimerization compared with pre-light. This suggests that more cooperative interactions of Ca_V1.2 channels can increase inotropy and thereby tune EC coupling.

The finding that physical association between channels could enhance their activity and amplify whole cell Ca²⁺ influx and inotropy was exciting, but even more so were our findings when we used this system to dimerize WT Ca_V1.2 with Ca_V1.2(G436R) TS mutant channels (187). As discussed above, the G436R (or G406R in humans) is a gain-of-function mutation that effects reduced inactivation and increased P_o in TS channels (71). When we fused WT channels to TS channels, somewhat unexpectedly, the WT channels exhibited some TS channel properties. This was uncovered by using a dihydropyridine-insensitive WT (T1066Y) channel so that after blue light illumination the individual contribution of WT channels to I_Ca could be dissected with application of nifedipine (11). We found that WT channels seemed to be strongly influenced by the higher-P_o TS channels and actually began to exhibit some deficits in their inactivation in a TS-like manner. Since TS is a heterozygous condition, this finding suggested that having just one high-P_o TS channel within a cluster of otherwise WT cooperatively gating channels could make them all behave abnormally and produce an arrhythmogenic substrate. Indeed, biolistic transfection of rat ventricular myocytes with WT-GI and TS-FKF1 channels frequently resulted in arrhythmogenic spontaneous Ca²⁺ release events and Ca²⁺ alternans after light-induced channel fusion (187). A later in silico modeling study concluded that aberrant levels of cooperatively gating channels could feasibly prolong action potential (AP) duration and produce Ca²⁺ alternans (188). TS is a long QT-producing condition that causes AP prolongation and life-threatening arrhythmias. This computational study revealed that Ca_V1.2 channel cooperativity could play a pivotal role in this pathology.

The observation that channels with the highest P_o can exert an influence on the activity of other neighboring channels within a cluster has implications for other modes of channel modulation, for example channel phosphorylation, which is known to produce enhanced P_o of Ca_V1.2 channels in heart, smooth muscle, and neurons. In cardiac muscle, PKA-mediated phosphorylation of Ca_V1.2 channel complexes has explicitly been linked to enhanced cooperativity of Ca_V1.2 channels and amplification of Ca²⁺ influx downstream of β-adrenergic receptor activation (50). In isolated mouse ventricular myocytes, isoproterenol (ISO)-mediated stimulation leads to increased insertion of Ca_V1.2 from subsarcolemmal reservoirs of endosome-localized channels (51), producing enhanced sarcolemmal expression of Ca_V1.2, and results in the appearance of large superclusters of Ca_V1.2 channels in the t-tubule membrane (50, 51). The “superclustering” response depended on PKA activation, as inhibitors of PKA (H-89 and PKAi) prevented their formation (50). Enhanced protein-protein interactions between adjacent Ca_V1.2 channels were visualized by bimolecular fluorescence complementation. This enhanced physical association facilitates cooperative allosteric interactions between channels, and, accordingly, examination of Ca_V1.2 single-channel activity in both Ca²⁺ sparklet and cell-attached patch recordings revealed enhanced cooperative gating of Ca_V1.2 in response to the β-agonist ISO. Consistent with the idea that cooperative gating is instigated by the highest-P_o channel in the cluster, these findings suggest that just a few phosphorylated Ca_V1.2 channels could have a disproportionately large effect on I_Ca and add another layer of complexity to β-adrenergic regulation of cardiac Ca_V1.2.

We have subsequently discovered that Ca²⁺ and Ca²⁺·CaM binding to the pre-IQ motif of Ca_V1.2 channels plays a key role in mediating the physical interaction between adjacent channels within a cluster (42) (FIGURE 4A). Whether Ca²⁺·CaM acts as a physical bridge between adjacent channels via pre-IQ to pre-IQ interactions is unknown, but a crystallography study supports this model (189). In this structural study, a dimer of Ca_V1.2 channel COOH-terminal tail segments was found to be bridged by two Ca²⁺·CaM molecules bound to the pre-IQ motifs. Debate around the validity of this model was sparked by a study suggesting that this dimer had no physiological relevance, noting that channels in Xenopus oocyte cell membrane exclusively expressed as monomers (190). However, Xenopus oocytes are quite distinct from mammalian somatic cells, as they are haploid. Although they have been an extremely useful tool for classical ion channel biophysicists, it is possible that they could lack endogenous expression of a vital element or scaffold that permits channel oligomerization, for example, AKAP5. Furthermore, we have successfully identified oligomeric channel assemblies in mammalian cell lines and in ventricular myocytes with stepwise photobleaching approaches and superresolution microscopy (29, 42, 50), so this may suggest that Xenopus oocytes are not the best platform for studies of mammalian Ca_V1.2 channel stoichiometry and assembly.

Our current model of Ca_V1.2 channel interactions has also been informed by FRET experiments that revealed that enhanced associations between COOH-terminal tails were formed dynamically as depolarization-stimulated Ca²⁺ influx occurred through these voltage-dependent channels (42). With this approach, a subpopulation of Ca_V1.2 channels were observed to remain closely associated with one another even after the Ca²⁺ influx that fueled the interactions had subsided because of voltage- and calcium-dependent channel inactivation. Similar results were observed after uncaging Ca²⁺. We hypothesize that this subpopulation of channels that remain associated with one another even after Ca²⁺ influx has ceased play a role in facilitation of I_Ca with successive depolarizing stimuli. This could be important during high-frequency stimulations, for example, during the fight-or-flight response when heart rate is elevated and APs arrive at the ventricular myocardium in more rapid succession than at resting heart rate. We tested this idea with paired-pulse facilitation experiments, and those data were well fit by a model that incorporated the channel recovery from inactivation and the channel “uncoupling” or cluster disassembly, which was informed by the decay kinetics of the FRET signal. Altogether, these results suggest that the degree of facilitation of I_Ca during high-frequency stimulation is directly proportional to the number of functionally interacting channels.

Recent work has shown that the number of channels in the t-tubule sarcolemma is increased by ∼40% during β-adrenergic receptor stimulation via enhanced sarcolemmal insertion of often intact channel clusters that originate in endosomal channel reservoirs and provide a functional reserve via triggered Rab4a and Rab11a recycling pathways (51). This enhanced channel expression and clustering at the membrane during a fight-or-flight response could cause an even greater degree of facilitation than that provided by the higher-frequency stimulation alone and could efficiently boost inotropy to meet the enhanced hemodynamic and metabolic demands of the body during acutely stressful events. Electrophysiological and imaging studies suggest that this stimulated channel-insertion response is an essential component of β-adrenergic regulation of Ca_V1.2, as disruption of the cytoskeletal delivery pathways prevented stimulated insertion and superclustering and eliminated the normal I_Ca augmentation response to ISO and the characteristic leftward shift that is a signature of PKA regulation of cardiac Ca_V1.2 channels (51).

How this stimulated insertion pathway elides with the recent finding from Steven Marx’s group that β-adrenergic regulation of Ca_V1.2 channels occurs as a result of disinhibition upon dissociation of phosphorylated Rad from the Ca_Vβ subunit of the channel complex (191) remains to be seen. There have been previous reports that RGK proteins including Rad interfere with Ca_V channel trafficking and reduce I_Ca by limiting its expression at the plasma membrane (192, 193). Furthermore, this reduction in channel trafficking is thought to be relieved by Rad phosphorylation and regulated by CaM and 14-3-3 (192). The impact of this on stimulated channel insertion is something that should be disentangled as these two hypotheses are assimilated. Our current understanding of Ca_V1.2 channel cooperativity is represented in FIGURE 4B.

A recent study has independently corroborated the findings that Ca_V1.2 channels gate coordinately (194). With the use of scanning patch clamp to record single-channel activity in defined sarcolemmal locations, a strikingly increase in Ca_V1.2 cooperativity was observed in failing rat ventricular myocyte t-tubule membranes consequent to myocardial infarction (MI). Examination of the all-points histogram reveals that the predominant single-channel openings of control myocytes were almost completely absent from post-MI t-tubule-localized channels, and in their place were doublet and quadruplet channel openings. This overt cooperative gating behavior was extinguished by inhibition of PKA (50). This finding is consonant with our previous report that β-adrenergic receptor stimulation enhanced Ca_V1.2 channel expression in the t-tubule membranes (51), leading to enhanced channel superclustering and increased cooperative gating and resulting in amplification of I_Ca (50). Thus, Ca_V1.2 cooperative gating seems to be a general property of these channels with broad physiological and pathological implications. We foresee that Ca_V1.2 cooperative gating may be observed in other tissues such as neurons, endocrine cells, and visceral smooth muscle.

Environmental pollutants, such as secondhand smoke, have also been shown to increase Ca_V1.2 cooperative gating in vascular smooth muscle, and this was associated with elevated myogenic tone of mesenteric arteries (195). Moreover, cooperative channel gating has been implicated in the pathology of hyperglycemia and diabetes. For example, in vascular smooth muscle from humans and mice, high glucose is thought to stimulate a particular pool of PKA that localizes to AKAP5-anchored complexes adjacent to Ca_V1.2 channels, causing Ca_Vα_1C phosphorylation on the COOH-terminal tail at serine residue 1928 (S1928) (45, 46, 52, 155). The consequence of this targeted phosphorylation is enhanced I_Ca current density and increased myogenic tone. Mice with a nonphosphorylatable S1928A substitution failed to display increased I_Ca and vasoconstriction responses to high-glucose or high-fat diet metabolic challenges, indicating that this phosphorylation event is required for the enhanced Ca_V1.2 channel activity and vascular reactivity that likely contributes to the enhanced risk of stroke, hypertension, and heart disease in diabetic patients. So, does cooperative gating of Ca_V1.2 factor into this enhanced activity? There is some evidence that it does. A study found that hyperglycemia and diabetes produced “persistent” Ca²⁺ sparklet activity in mouse cerebral artery smooth muscle that could be abolished by PKA inhibitors (172). Close examination of the Ca²⁺ sparklet recordings from that study shows the hallmarks of cooperative gating behavior including frequent time-correlated openings of multiple Ca_V1.2 channels and indeed closings. Although the link to cooperativity has not been explicitly stated, it appears to play a role in this pathological response. Future studies should examine this interesting possibility. Moreover, studies should determine whether S1928 phosphorylation plays a role in modulating cooperative gating of Ca_V1.2 channels in vascular smooth muscle during physiological conditions and in diabetic hyperglycemia. It is tempting to speculate that S1928 phosphorylation may be a key factor underlying cooperative gating of Ca_V1.2 channels in other cell types such as neurons (196, 197).

3.6. Ca_V1.3 Channel Structure and Function

The pore-forming subunit of Ca_V1.3 is Ca_Vα_1D encoded by cacna1d (75). Ca_V1.3 and Ca_V1.2 are expressed in many of the same tissues, including brain, heart, and endocrine glands (68, 198). Ca_V1.3 channel inactivation kinetics are slower than Ca_V1.2, and they display a more left-shifted voltage dependence of activation and begin to open at more hyperpolarized potentials of approximately −60 mV, compared with the approximately −40 mV activation threshold of Ca_V1.2 (199). However, when recording from cells that have varying proportions of Ca_V1.2 and 1.3 channels the value falls somewhere in between, and in the absence of a specific pharmacological blocker that can distinguish between these two L-type channels the physiological role of Ca_V1.3 was unresolved for many years. The generation of a transgenic Ca_V1.3−/− mouse exposed the functional importance of these channels (reviewed in Ref. 198). Accordingly, the knockout mice displayed two striking phenotypes. First, they were deaf, confirming the critical role of Ca_V1.3 in the auditory system, where they are expressed on the inner (IHC) and outer hair cells of the cochlea (200). Second, they were bradycardic, a telling consequence of the lack of Ca_V1.3 in pacemaking sinoatrial node (SAN) and atrioventricular node (AVN) cells, where they play a role in the diastolic depolarization phase of the nodal action potential (200, 201).

Hyperpolarization-activated cyclic nucleotide-gated (HCN) channels HCN1, HCN2, and HCN4 that underlie the funny current (I_f) are also famously critical players in diastolic depolarization, allowing these Ca_V1.3−/− cells to display automaticity (202) (sect. 9). Thus, Ca_V1.3 knockout is not lethal because of a total lack of pacemaker activity, but SAN dysfunction is certainly apparent in these animals. Indeed, telemetric recordings from Ca_V1.3−/− mice, revealed electrocardiograms with high R-R interval variability and sinoatrial bradycardia at low resting heart rates that could be rescued by treatments that increased heart rate, e.g., exercise, or injection of ISO (a β-agonist that mimics sympathetic stimulation) or atropine (a muscarinic receptor blocker that dampens parasympathetic effects) (200). The relatively hyperpolarized threshold of activation of Ca_V1.3 is thought to contribute to its role in diastolic depolarization in nodal cells, and the more depolarized activation threshold of Ca_V1.2 explains its inability to compensate for Ca_V1.3 absence in the SAN and AVN of knockout animals (203).

Ca_V1.3 channels are also involved in shaping the firing patterns and modulating the resting potentials of neurons in the central nervous system (CNS) (117, 204). In the brain, Ca_V1.2 channels outnumber Ca_V1.3 by ∼9:1 and they are often both expressed within the same neuron (205, 206). In the dopaminergic neurons in the substantia nigra, Ca_V1.3 channels have been implicated in the pathophysiology of Parkinson’s disease (PD), where they orchestrate the rhythmic pacemaking of these cells (207–210).

Splice variants of Ca_V1.3 include the so-called “long” and “short” isoforms that result from alternative splicing of Ca_Vα_1D exon 42 and differ in the length of their COOH-terminal tail, their inactivation kinetics, and their threshold of activation (211). The long Ca_V1.3_L isoform (also known as Ca_V1.3₄₂), has a 695-amino acid COOH-terminal tail containing a COOH-terminal modulatory domain (CTM). Within this domain, interactions between a proximal COOH-terminal regulatory domain (PCRD) and a distal COOH-terminal regulatory domain (PCRD) strongly reduce Ca²⁺·CaM binding to the IQ motif, resulting in reduced Ca²⁺-dependent inactivation of these channels (211). The CTM is absent from the short splice variant Ca_V1.3_S (also known as Ca_V1.3_42A), which has a 183-amino acid COOH-terminal tail that offers less steric hindrance to Ca²⁺·CaM-IQ motif interactions. Thus Ca_V1.3_S inactivates more rapidly than Ca_V1.3_L (211). Despite the enhanced inactivation that limits Ca²⁺ influx, Ca_V1.3_S channels display ∼2.5-fold larger I_Ca densities than Ca_V1.3_L channels in the absence of any detectable changes in expression levels (211) or unitary current amplitude (12, 212). The current-amplifying effects of positive cooperative gating, exclusively of Ca_V1.3_S and not Ca_V1.3_L, could explain this discrepancy in I_Ca. Indeed, our group has reported that Ca_V1.3_S channels undergo Ca²⁺·CaM-dependent cooperative interactions in tsA-201 cells and hippocampal neurons as discussed in detail in sect. 3.7 (12).

3.7. Cooperativity Among Ca_V1.3 Channels

Almost a decade after our first study demonstrating cooperativity of Ca_V1.2 channels (26), we investigated whether a similar phenomenon could occur for Ca_V1.3 channels (12). As noted above, it had already been reported that Ca_V1.3_S channels generated larger I_Ca density than Ca_V1.3_L channels, which could not be explained by changes in channel expression or unitary current, and so we started there. In whole cell patch-clamp experiments, we noticed that I_Ca in Ca_V1.3_L-expressing tsA-201 cells was ∼40% smaller than I_Ba. This was not unexpected given that these channels, like all voltage-gated Ca²⁺ channels, are more permeable to Ba²⁺ than Ca²⁺ (213). However, Ca_V1.3_s only exhibited an ∼15% reduction in current when the extracellular solution was switched from one containing Ba²⁺ to one containing Ca²⁺. This finding suggested that Ca²⁺ might alter the activity of Ca_V1.3_S. We tested this hypothesis in cell-attached patch single-channel recordings and found that NP_o (where N represents channel number) in Ca_V1.3_s-expressing cells was ∼1.5-fold higher with Ca²⁺ as the charge carrier than with Ba²⁺, whereas no such Ca²⁺-dependent increase in NP_o was observed in Ca_V1.3_L-expressing cells. This was reminiscent of the Ca²⁺ dependence of Ca_V1.2 cooperativity we had reported previously, and accordingly we examined Ca_V1.3 Ca²⁺ sparklets for the two different isoforms, finding that Ca_V1.3_L channels displayed more overt cooperative gating behavior and had correspondingly higher coupling coefficients than Ca_V1.3_L channels that appeared to largely gate independently. Superresolution microscopy confirmed that both Ca_V1.3_L and Ca_V1.3_S channels cluster in tsA-201 cells, suggesting that although clustering is necessary for cooperative channel interactions, it is not sufficient. Ca_V1.3 channels were also observed to cluster in cultured rat hippocampal neurons, where stepwise photobleaching of neurons transfected with Ca_V1.3_S-EGFP revealed that clusters of Ca_V1.3 contained eight channels on average. Consistent with the hypothesis that only the Ca_V1.3_S slice variant undergoes cooperative gating, light-induced dimerization of these channels produced a 1.35-fold amplification in their I_Ca, whereas similarly evoked dimerization of Ca_V1.3_L channels did not alter I_Ca. As was the case for Ca_V1.2 channels, spontaneous physical interactions of Ca_V1.3_S COOH-terminal tails occurred in a manner dependent on Ca²⁺·CaM binding to the pre-IQ motif. The functional consequences of Ca_V1.3 channel cooperativity in neurons were assessed by recording action potentials from neurons transfected with Ca_V1.3_S-CIBN and Ca_V1.3_S-CRY2 before and after blue light-induced dimerization. Strikingly, channel fusion was observed to enhance neural excitability and firing rate. It is therefore quite reasonable to posit that cooperative gating of Ca_V1.3_S makes an important contribution to the sustained firing observed in motoneurons and other cells in the CNS (214). A recent publication has shown robust expression of Ca_V1.3_S mRNA in the substantia nigra pars compacta in the 1-methyl-4-phenyl-1,2,3,6-tetrahydropyridine (MPTP) mouse model of Parkinson’s disease (207), although the physiological role of Ca_V1.3 cooperativity in cardiac pacemaking remains an important open question.

4. TRANSIENT RECEPTOR POTENTIAL CHANNELS

The transient receptor potential (TRP) channels are a superfamily of ion channels encoded by 28 different genes (215, 216). Sequence homology analyses have revealed that TRP channels can be further classified into six different subfamilies, namely canonical TRP (TRPC), vanilloid TRP (TRPV), melastatin TRP (TRPM), polycystin TRP (TRPP), ankyrin TRP (TRPA), and mucolytic TRP (TRPML) channels (215, 216). Each TRP channel subunit consists of six membrane-spanning helices (S1−S6) (FIGURE 5A) (216). The amino and carboxy terminals of each subunit are located intracellularly and contain several distinctive domains that regulate channel function (217). Several published cryo-EM structures of TRP channels (TRPV1, TRPV2, and TRPA1) at near-atomic resolution support the idea that TRP channels are formed by the assembly of one or more subunits into tetramers around a central pore formed by the S5 and S6 helices (216). The selectivity filter is located in the upper region of the pore and is modulated by endogenous and exogenous ligands (216). Given the ubiquitous expression of a diverse number of TRP channels in a given cell, it is likely that they are formed by homo- and/or heterotetramers.

FIGURE 5.

Model of cooperative gating of transient receptor potential (TRP) channels. A: TRP subunit topology. PIP₂, phosphatidylinositol 4,5-bisphosphatase. B: model by which G_q agonists trigger AKAP5-dependent, PKC-mediated cooperative gating of TRPV4 channels. G_qPCR, G_q protein-coupled receptor. Figure created with Biorender.com.

TRP channels are nonselective cation-permeable channels with permeability ratios (P_Ca/P_Na) in the range of 0.1–20 (216). Although most TRP channels are permeable to Ca²⁺, TRPM4 and TRPM5 currents are carried almost exclusively by Na⁺ ions, with minimal if any contribution made by Ca²⁺ (215, 216). TRP channels can be activated by numerous intracellular and extracellular stimuli including changes in temperature, different Ca²⁺ signals, mechanical stress, and chemical cues. TRP channels can act as integrators of different signaling pathways to elicit specific cellular responses (215, 216). Indeed, TRP channels modulate membrane potential, intracellular Ca²⁺ signaling, generation of action potentials, migration, proliferation, and many other processes. The physiological processes mediated by TRP channel activation include pain and taste sensation, control of vascular tone, immunological responses, and cellular homeostasis (215, 216). Dysfunction of TRP channels leads to several channelopathies, including cardiovascular complications due to gain- and loss-of-function mutations, renal diseases, blindness, atrophy of several tissues, and pain disorders. These pathologies have been extensively discussed and documented in recent comprehensive review papers on TRP channels (215, 216). In sects. 4.1–4.4, we examine examples of studies describing cooperative gating of TRP channels as well as the current understanding of mechanisms underlying this gating modality and its functional implications, topics that have not been previously reviewed.

4.1. TRP Channels Undergo Cooperative Gating

The discovery that TRP channels can engage in cooperative gating was contingent on optical recording techniques, including TIRF and spinning disk confocal imaging (14, 17, 25) that enabled the detection of Ca²⁺ influx via single or clusters of Ca²⁺-permeable channels with a high spatial and temporal resolution over a large surface area. The first example of cooperative gating of TRP channels was described for TRPV4 in native mesenteric endothelial cells with an en face preparation from a mouse expressing the GCaMP2 Ca²⁺ sensor specifically in endothelial cells (166). About the same time, an independent study using pressurized intact cremaster arterioles loaded with the Ca²⁺ indicator fluo4-AM described the induction of endothelial events produced by the simultaneous opening of two or more TRPV4 channels (218). Subsequently, TRPV4 cooperative gating has been described in cultured human endothelial cells and cultured human airway smooth muscle cells loaded with fluo4-AM (167, 219), native mouse cerebral and mesenteric smooth muscle cells loaded with the fast Ca²⁺ indicator fluo-5F (43, 44), native mouse and human mesenteric endothelial cells from endothelium-specific GCaMP2 mice or cells loaded with fluo4-AM (220, 221), and native mouse pulmonary endothelial cells from endothelium-specific GCaMP2 mice (222). Intriguingly, TRPV4 activity and the frequency of cooperative gating differ with biological sex and vascular bed (43). TRPV4 activity and cooperative gating is higher in male than female smooth muscle cells. Moreover, TRPV4 activity and cooperative gating are of equal magnitude in cerebral pial and parenchymal smooth muscle cells but completely absent in mesenteric smooth muscle from both sexes. These results highlight important sex-specific and vessel-dependent regulation of TRPV4 biophysical properties that may be general properties of other TRP channels.

Although less extensively studied, there are reports of cooperative gating of TRPA1 channels in mouse cerebral endothelial cells from endothelium-specific GCaMP6f mice or cells loaded with fluo4-AM (168, 170) and TRPV3 in cultured mouse endothelial cells loaded with fluo4-AM (169). General Ca²⁺ signaling properties of the TRP-mediated events reported in these studies are somewhat different, which is likely due to intrinsic differences in channel activity for different cell types and/or to the use of different acquisition approaches. In TABLE 1, we provide a summary of the unitary current amplitude, number of channels per membrane site, as well as the frequency, amplitude, and spatial spread of the Ca²⁺ signals produced by different TRP channels in multiple cells. All reports, however, indicate that cooperative gating of TRP channels may result from a two- to four-channel metastructure. These results suggest that cooperative gating appears to be a general property of TRP channels arising from aggregates of a given number of channels.

Table 1.

Properties of Ca²⁺ signals mediated by TRP channels

TRP Channel (reference)	Cell Type	Unitary Amplitude	Maximum Number Channels/Site	Open Time/Po	Spatial Spread	Event Frequency
TRPV4 (219)	Human cultured microvascular EC	0.39 ± 0.01 (ΔF)	∼5	520 ± 40 ms	6.26 ± 2.01 μm2	0.11 ± 0.02 Hz
TRPV4 (167)	Human cultured airway SMC	0.06 (ΔF)	∼3	110–250 ms	0.7 μm2	0.2 ± 0.5 Hz
TRPV4 (44)	tsA-201 cells	∼50 nM	∼1
TRPV4 (44)	Native mouse SMC	∼48 nM	∼4	12–68 ms		2–5 sites/cell
TRPV4 (166, 221)	Native mesenteric EC	0.19 (ΔF)	∼4	37.0 ± 0.7 ms	11.2 ± 0.4 μm2	10–25 sites/field
TRPV4 (222)	Native pulmonary EC	0.29 (ΔF)	∼4	0.03–0.16 (Po)	∼20 μm2	5–30 sites/field
TRPV3 (169)	Primary cerebral EC	∼0.2 (ΔF)	Unknown	70–170 ms	0.7 μm2	0.05–0.4 Hz
TRPA1 (168)	Native cerebral EC	∼0.25 (ΔF)	∼3	96 ms	5.02 ± 0.60 μm2	1–8 sites/field

EC, endothelial cell; P_o, open probability; SMC, smooth muscle cell; TRP, transient receptor potential; TRPA, ankyrin TRP; TRPV, vanilloid TRP; ▵F, change in fluorescence from the calcium indicator.

4.2. Mechanisms of Cooperative Gating of TRP Channels

TRP channel cooperative gating is predicated on the close proximity of two or more channels and is likely to be supported by an association with a structural protein. Accordingly, there has been a substantial effort to understand the role of the scaffold protein AKAP5 in TRPV4 clustering (5, 43, 44, 221, 223). The focus on AKAP5 stems from work suggesting its association with TRPV4 channels (224) and its role in regulating the extent of cooperative gating of L-type Ca²⁺ channels (13).

On first principles, the clustering of ion channels must be intimately related to their trafficking properties. The mechanisms underlying the subcellular dynamics and translocation of TRP channels, including TRPV4, have been extensively reviewed (225, 226). However, an emerging concept is that dynamic trafficking of ion channels is essential for sustaining their clustering and cooperative gating (50, 51, 227). By combining experimental and in silico approaches, a model for TRPV4 clustering has been proposed in which channel clustering is mediated by a stochastic self-assembly process (5). This hypothesis is supported by superresolution data showing that TRPV4 cluster size has an exponential distribution indicative of a Poisson process. A similar exponential distribution of cluster sizes has been observed for TRPML1 channels in vascular smooth muscle (228), suggesting that this spatial arrangement is likely to be a general feature of TRP channels. Sato et al. (5) also found that TRPV4 cluster size and density seem to be regulated by a feedback mechanism that controls the size and the number of clusters at the surface membrane. This proposal rests on superresolution data showing that TRPV4 clusters rapidly increased in size and density after transfection in HEK cells but reached a steady state after 24 h. Intriguingly, the accompanying in silico model predicted that TRPV4 channel and cluster dwell times at the membrane would be in the range of minutes (5). This suggests that the insertion and removal of TRPV4 channels and clusters must be dynamic and well controlled to sustain the basal TRPV4 current, as observed for other TRP channels (229–231), and are likely to engender cooperative channel gating. A similar clustering formation mechanism accounts for Ca_V1.2, Ca_V1.3, and large-conductance Ca²⁺-activated K⁺ (BK_Ca) channel cluster size in different cells (5). These results suggest a potentially general mechanism for ion channel (including TRP channels) cluster formation, which may influence cooperative gating. This model requires further evaluation.

Existing evidence suggests that elevations in intracellular Ca²⁺ and AKAP5 contribute to cooperative gating of TRPV4 in mesenteric endothelial and cerebral smooth muscle cells (43, 44, 221, 223). The molecular details underlying the AKAP5-TRPV4 interaction that could sustain cooperative gating are unresolved but may include protein-protein contacts between the carboxy terminals of each protein (224). However, genetic ablation of AKAP5 does not completely abolish the induction of cooperative gating of TRPV4 channels in mesenteric endothelial and cerebral smooth muscle cells (43, 44, 221, 223). Thus, there may be other structural proteins (e.g., α-actinin) that facilitate cooperative gating of TRPV4 channels in the absence of AKAP5, a conjecture that requires experimental confirmation. Furthermore, AKAP5 may be playing a more traditional role for this scaffold in targeting signaling proteins (e.g., PKC, PKA, PP2B) to modulate TRPV4 channel activity and local intracellular Ca²⁺ concentration that promotes TRPV4 cooperative gating.

Supporting the aforementioned possibility, superresolution imaging in cerebral and mesenteric vascular smooth muscle revealed TRPV4 and AKAP5 clusters of various sizes (43). The association between TRPV4 and AKAP5 clusters is augmented in response to acute ANG II exposure in pial and parenchymal smooth muscle cells (43). This treatment appears to facilitate AKAP5-anchored PKC-dependent stimulation of TRPV4 channel activity and may account for the increased frequency of cooperative gating events in these cells. Conversely, chronic ANG II signaling activation, as during hypertension, promotes an increase in the distance between TRPV4 and AKAP5 clusters that results in a reduction of both channel open probability and the frequency of cooperative events in TRPV4 channels. Intriguingly, the chronic ANG II increase in TRPV4-AKAP5 intermolecular distance in pial and parenchymal smooth muscle is comparable to the basal distance between these proteins in mesenteric cells (43), which is in line with the vessel-specific regulation of TRPV4 channels described above. Thus, the available evidence suggests that TRPV4 activity and cooperative gating are highly dependent on the distance between TRPV4 channel clusters and AKAP5. Accordingly, AKAP5 may compartmentalize TRPV4 and PKC in close proximity to each other to regulate the channel biophysical properties. In agreement with this idea, dialysis of an activated PKC in mesenteric smooth muscle increased TRPV4 current density and likely the frequency of cooperative events (43). These results suggest that regulation of TRPV4 channel function is limited by the availability of AKAP5-anchored PKC near the channel.

On the basis of results described above, we proposed the following model for cooperative gating of TRPV4 channels (FIGURE 5B). In this model, AKAP5, PKC, and TRPV4 clusters form a nanocomplex via a self-assembly, feedback-regulated mechanism. Stimulation of AKAP5-anchored PKC promotes the activity of nearby TRPV4 channels. The resultant increase in Ca²⁺ influx through a TRPV4 channel primes the other channels within the nanocomplex to undergo as yet undetermined structural rearrangements that facilitate cooperative gating. Considering that PKA increases TRPV4 current density (232), the aforementioned model may have broad implications for regulation of TRPV4 and other TRP channel function and cooperative gating by different G protein-coupled receptors. Because TRPV4 channel activity is modulated by phosphatidylinositol 4,5-bisphosphate (PIP₂) (233–236), it would be interesting to determine whether coupled gating of this channel during ANG II/PKC/Ca²⁺ signaling activation requires PIP₂.

4.3. Physiological and Pathological Implications of TRP Cooperative Gating

Cooperative gating of TRP channels amplifies Na⁺ and Ca²⁺ influx (15, 23) that can influence membrane potential and intracellular Ca²⁺-dependent processes. Perhaps not surprisingly, the frequency of cooperative gating of TRP channels is low regardless of the TRP isoform and the cell type (43, 44, 166–170, 219, 221–223). Yet different stimuli can promote cooperative gating of TRP channels. For instance, selective agonists for TRPV4 [e.g., GSK1016790A (43, 44, 167, 219, 221–223)] and TRPA1 [e.g., AITC or 4-HNE (170, 228)] have been shown to increase the frequency of cooperative events. Activation of G_q protein-coupled receptors upstream of PKC with carbachol or ANG II also stimulates TRPV4 cooperative gating (43, 44, 166, 223). The dietary compound carvacrol stimulates TRPV3 cooperative gating (169). Moreover, low pressure was shown to stimulate TRPV4 signals that seem to be produced by the opening of two or more channels, whereas high-pressure conditions prevented/blocked these signals (218).

The increased cooperative gating of TRPV4, TRPV3, and TRPA1 channels can be correlated with endothelium-dependent vasodilation mechanisms in cerebral, mesenteric, cremaster, and pulmonary arteries (166, 168–170, 218, 221, 222). These studies suggest that an increase in cooperative gating of mesenteric and cerebral endothelial TRPV4, TRPV3, and TRPA1 at the myoendothelial projections could amplify local Ca²⁺ near intermediate-conductance (IK) and small-conductance (SK) Ca²⁺-activated K⁺ channels. Activation of IK/SK channels in endothelial cells causes an efflux of K⁺ ions and hyperpolarization of the plasma membrane of endothelial cells (237, 238). The K⁺ ions could activate nearby inward-rectifying K⁺ channels in smooth muscle cells to cause relaxation (149, 239). In addition, endothelial cell hyperpolarization can be transmitted to the adjacent smooth muscle cells via gap junctions at myoendothelial projections to produce relaxation (237, 238). These mechanisms may act in concert to promote vasodilation. In small pulmonary arteries, however, cooperative gating of TRPV4 channels in endothelial cells has been linked with Ca²⁺-dependent activation of the endothelial nitric oxide synthase (eNOS)/nitric oxide (NO) signaling pathway (222). NO can then diffuse into adjacent smooth muscle cells to activate the guanylyl cyclase/protein kinase G (PKG) pathway, leading to smooth muscle relaxation and vasodilation. Overall, results in mesenteric/cerebral versus pulmonary preparations suggest distinct mechanisms by which cooperative gating of TRP channels may contribute to regulating arterial reactivity.

In smooth muscle, cooperative gating of TRPV4 channels may promote Ca²⁺ sparks (e.g., Ca²⁺ release via ryanodine receptors) in the sarcoplasmic reticulum (240). Ca²⁺ sparks can then trigger spontaneous transient outward currents (STOCs) via activation of large-conductance Ca²⁺-activated K⁺ (BK_Ca) channels, leading to vascular smooth muscle relaxation (149, 241). Consistent with this possibility, it was found that TRPV4 channels oppose ANG II-induced vasoconstriction (44), although the contributions of endothelial versus smooth muscle TRPV4 were not differentiated in this study. Cooperative gating of TRPV4 may also be correlated with activation of the calcineurin/nuclear factor of activated t cell (NFAT) transcription factor in airway smooth muscle as well as proliferation of these cells (167). Thus, cooperative gating of TRP channels may have broad and significant physiological impact on a number of intracellular signaling cascades.

Cooperative gating of TRP channels may play a key role in mediating several pathological conditions. For example, it has been reported that hypoxic conditions stimulate endothelial TRPA1 channel activity, and likely their cooperative gating, to promote vasodilation of cerebral arteries (170). This study also found that hypoxia-induced endothelial TRPA1 activation may be protective against ischemic stroke (170). Cooperative gating of TRPV4 channels in airway smooth muscle may contribute to activation of the calcineurin/NFAT signaling pathway (167). This may lead to airway smooth muscle proliferation that could underlie some of the remodeling observed during asthma. Thus, targeting TRPV4 or preventing TRPV4 cooperative gating might provide novel therapeutic strategies for asthma treatment.

In vascular smooth muscle, TRPV4 activity and cooperative gating are reduced in male cerebral cells in an angiotensin II-induced hypertension mouse model (43). The effects of hypertension on TRPV4 channels properties were correlated with a reduction in the cluster density of TRPV4 channels and AKAP5, as well as an increase in the intermolecular distance between these proteins as determined with superresolution imaging. Intriguingly, hypertensive female vascular smooth muscle from pial and parenchymal branches showed a range of distances between TRPV4 and AKAP5, with all of them exceeding a 200-nm radius (43). This is critical because the overall level of TRPV4 activity, and the likely induction of cooperative channel gating, is determined by the relative distance between the channel and AKAP5, which seems to have a threshold for regulation at ∼200 nm (43). The disruption of vascular smooth muscle TRPV4-AKAP5 signaling during hypertension may prevent optimal activation of Ca²⁺ sparks and STOCs, resulting in impaired vasodilation (44, 149). Reduced TRPV4 activity and cooperative gating has also been observed in mesenteric endothelial cells from ANG II-induced hypertensive mice (223). As in smooth muscle, the decreased TRPV4 activity and cooperative gating during hypertension was correlated with impaired AKAP5 signaling. Specifically, it was found that AKAP5 disappears from the myoendothelial projections where it facilitates PKC dependent modulation of TRPV4 channels. The absence of AKAP5 at myoendothelial projections during hypertension is likely due to reduced expression of the scaffold rather than its redistribution, as changes in AKAP5-associated signal were not apparent in other endothelial subcellular domains. Disruption of the AKAP5-PKC-TRPV4 axis in endothelial cells during hypertension prevented optimal signaling of TRPV4 Ca²⁺ to IK/SK channels, leading to impaired transmission of endothelium-dependent hyperpolarization and vasodilation.

A reduction in TRPV4 activity and cooperative gating was also observed in mesenteric endothelial cells from a high-fat diet model of obesity (221). This reduction was correlated with impaired AKAP5-anchored PKC signaling to TRPV4 channels but not with changes in the expression of TRPV4 or AKAP5 or with changes in the association between these proteins. Altered AKAP5-PKC signaling to TRPV4 seems to be mediated by an increase in local peroxynitrite production due to elevated activity of the inducible nitric oxide synthase (iNOS) and the NADPH oxidase 1 (NOX1). It is suspected that increased peroxynitrite oxidizes AKAP5 at cysteine 36 to prevent the interaction of the scaffold with PKC and its subsequent downstream signaling to TRPV4 that will promote cooperative channel gating. These alterations in the AKAP5-PKC-TRPV4 signaling axis by elevated peroxynitrite were found to impair vasodilation and contribute to increased blood pressure in obese mice.

On the basis of the above results, we propose that reduced endothelial and smooth muscle cell TRPV4 channel activity and cooperative gating may be key underlying factors contributing to and promoting vascular dysfunction during hypertension and obesity. Moreover, alterations in TRP channel activity and cooperative gating may have broad implications in modulating many cell and organ functions in health and disease, which remain to be completely elucidated.

4.4. TRP Clustering and Functional Coupling With Other Ion Channels

The coupling of TRP channels with several other ion channels is well documented in the literature (159, 228, 242–250). This cross-coupling between TRP and other ion channels regulates many processes in neuronal, cardiac, and vascular cells in health and disease (215, 251, 252), thus highlighting its significance. Basic understanding of the functional coupling between TRP and other ion channels has been advanced principally through superresolution microscopy that enables the visualization of these proteins at a nanoscale level.

For example, TRPV1 channels were found to cluster in Chinese hamster ovary (CHO) cells overexpressing the TRPV1 subunit and in sensory neurons (159). Using a sophisticated triple labeling approach, the authors found that TRPV1 could form supercomplexes within discrete nanometer regions with clusters of AKAP5 and KCNQ2 channels as well as clusters of AKAP5 and Ca_V1.2 channels in sensory neurons. These supercomplexes appear to be tethered together by AKAP5, as genetic ablation of this scaffold prevents the close association between TRPV1-KCNQ2 and TRPV1-Ca_V1.2 complexes in sensory neurons (FIGURE 6, A and B). These results suggest that AKAP5 may serve a dual role as 1) an organizer of signaling complexes to regulate protein function, as traditionally expected (253), and 2) a structural anchor supporting multi-ion channel supercluster formation.

FIGURE 6.

Functional coupling of transient receptor potential (TRP) channels with other ion channels. A: functional coupling between TRPV1 and KCNQ2 modulates neuronal excitability. PIP₂, phosphatidylinositol 4,5-bisphosphatase. B: clustering of TRPV1 and Ca_V1.2 channels regulates TRPV1 desensitization in neurons. C: a tripartite clustering of TRPV1, inositol trisphosphate (IP₃) receptor (IP₃R), and ANO1 controls membrane depolarization and dorsal root ganglion (DRG) excitability. PLC, phospholipase C. D: mucolytic TRP (TRPML)1s located in immobile lysosomes near ryanodine receptor (RyR) channels control large-conductance (BK) channel activity to regulate vascular smooth muscle contractility. P_o, open probability. STOCs, spontaneous transient outward currents. Figure created with Biorender.com.

In a clever series of experiments, it was found that these AKAP5-dependent multi-ion channel superclusters are functionally relevant (159). Activation of TRPV1 channels in sensory neurons led to an AKAP5-dependent inhibition of M currents (159). Although the specific mechanisms for TRPV1-mediated inhibition of M currents were not directly explored in this study, it was suggested that TRPV1 activation could deplete local levels of the phosphoinositide phosphatidylinositol 4,5-bisphosphate (PIP₂), likely within the AKAP5-dependent supercluster nanodomain. Because the KCNQ channel open state is highly sensitive to PIP₂ (254), the TRPV1-mediated depletion of PIP₂ results in suppression of M currents, leading to membrane depolarization and increase in neuronal excitability (FIGURE 6A).

On the other hand, it was also reported that TRPV1 channel activation may lead to activity-dependent desensitization of TRPV1 currents (159). Intriguingly, this was found to require engagement of the Ca_V1.2 gating machinery and expression of AKAP5. Accordingly, both nifedipine, a Ca_V1.2 inhibitor that blocks the gating machinery, and genetic ablation of AKAP5 significantly ameliorated activity-dependent TRPV1 desensitization. The mechanisms by which Ca_V1.2 may promote activity-dependent TRPV1 desensitization are currently unclear but may include local AKAP5-Ca_V1.2 Ca²⁺-dependent pathways, AKAP5-mediated physical coupling between both channels leading to conformational changes that alter TRPV1 biophysical properties, or a combination of both (FIGURE 6B).

In independent experiments using triple-labeling superresolution, it was also found that TRPV1 could form a supercomplex with clusters of the plasma membrane Ca²⁺-activated Cl⁻ channel ANO1 and the endoplasmic reticulum (ER) IP₃ receptor (IP₃R) in dorsal root ganglion (DRG) neurons (242). Whether AKAP5 contributes to the formation of the TRPV1-ANO1-IP₃R supercomplex was not examined, and the mechanisms by which the complex is assembled are unclear at this time. There are data that suggest that TRPV1 activation modulates ANO1 activity via two distinct and likely synergistic pathways (242). First, a local increase in Ca²⁺ influx via TRPV1 may stimulate ANO1 activity. Second, TRPV1 activation could promote phospholipase C (PLC) activity, leading to activation of IP₃Rs. Given the close physical proximity between IP₃Rs and ANO1, Ca²⁺ release via IP₃R can maximize ANO1 activity. These concurrent processes are likely to amplify Cl⁻ efflux and membrane depolarization in DRG neurons. The TRPV1-ANO1-IP₃R supercomplex may not require physical interaction between channels/receptors but rather a close association in which the local Ca²⁺ signal generated by TRPV1 and IP₃Rs can be sensed by ANO1 (FIGURE 6C).

In vascular smooth muscle, superresolution imaging revealed the presence of TRPML1 clusters in lysosomes that are in close association with ryanodine receptor (RyR) clusters located at the sarcoplasmic reticulum (228). Unexpectedly, TRPML1 channels were found in an immobile lysosomal pool that is thought to be in close proximity to RyRs. It is unclear whether a physical link occurs between TRPML1 and RyR. Nevertheless, this organization places TRPML1 in an ideal position to trigger activation of RyRs to produce Ca²⁺ sparks [i.e., Ca²⁺ release via RyR (127, 241)]. The resultant Ca²⁺ spark can then promote STOCs that affect vascular reactivity and blood pressure (FIGURE 6D).

Taken together, the results described above support the notion that functional, multi-ion channel supercomplexes that include one or several TRP channels are regularly formed in several different cell types. These supercomplexes can play key roles in regulating cellular function in health and disease. This assumption requires further exploration, thus representing a fertile area of research.

5. Na_V CHANNELS

The principal physiological function of Na_V channels is to regulate sodium ion flux across cell membranes (255). Channel activation results in membrane depolarization that generates the depolarizing phase of the action potential in excitable cells and facilitates the propagation of action potentials along contiguous cellular processes (1). Aberrations in Na_V channel function can exert profound effects on membrane excitability and may confer severe dysfunction in nerve and muscle tissues, including loss of sensation, chronic pain, cardiac failure, and paralysis (256).

The biophysical properties of Na_V channels have been extensively studied in neurons (e.g., Ref. 257) and heterologous expression systems (e.g., Ref. 258). Na_V channels are a ubiquitous class of transmembrane protein complexes composed of an α-subunit that forms the pore of the channel, which is structurally responsible for voltage sensing and gating as well as ion permeation, and one or more auxiliary β-subunits that modulate the voltage dependence, permeability, and kinetics of Na_V gating (259–261). In addition, the β-subunits are important regulators in cell adhesion, signal transduction, and channel expression at the plasma membrane (262).

The α-subunits of Na_V channels are large proteins of ∼250 kDa composed of four homologous domains (I–IV) consisting of six α-helical transmembrane-spanning segments (S1–S6) (FIGURE 7A) (263). The transmembrane segments are connected via small intracellular and extracellular loops, whereas homologous domains are connected through larger intracellular loops (264). Ten different α-subunit isoforms have been discovered to date that are expressed in different types of cells. Nine of the isoforms generate voltage-gated channels (Na_V1.1, Na_V1.2, Na_V1.3, Na_V1.4, Na_V1.5, Na_V1.6, Na_V1.7, Na_V1.8, and Na_V1.9). The tenth isoform (Na_VX) appears to reside within a related protein family, as it does not encode a voltage gate (265).

FIGURE 7.

Voltage-gated Na⁺ (Na_V) channel structure and a model of coupled Na_V channels. A: illustration of the pore-forming Na_V and auxiliary Na_Vβ subunits. Location of the 14-3-3 protein is highlighted in the I–II linker. B and C: models of the interaction between 2 Na_V channels based on structures by Jiang et al. (446) and Yang et al. (447) and data from Clatot et al. (10) and Allouis et al. (448). Figure created with Biorender.com.

Na_V1.1, Na_V1.2, Na_V1.3, and Na_V1.7 isoforms are principally expressed on the membranes of central and peripheral neurons. The Na_V 1.4 isoform is expressed predominantly in skeletal muscle cells, whereas the Na_V1.6 isoform is expressed in central neurons. Na_V1.5 is primarily expressed in the heart and Na_V1.8 and Na_V1.9 on neurons whose cell bodies reside in the dorsal root ganglia. Tetrodotoxin blocks ion permeation in all of the isoforms, though the binding affinities differ substantially among the isoforms (for review and references, see Ref. 266).

As mentioned above, the α-subunit of Na_V channels forms the channel’s pore and is responsible for its voltage sensitivity and selectivity. The S4 transmembrane segments of the α-subunit constitute the voltage sensor in Na_V channels, as is the case for other classes of voltage-gated ion channels (263). Depolarization of the membrane leads to a translocation of the positively charged S4 segments toward the outside of the membrane and initiates channel activation (267). The actual movement of the charges within the α-subunit, primarily confined to the intramembrane region of S4 segments, is proposed to produce the gating currents and channel opening (268).

The voltage sensor component of the α-subunit of Na_V channels is also associated with the voltage dependence of channel inactivation (269), which plays out on three distinct timescales: fast, slow, and ultraslow inactivation (reviewed in Ref. 270). Inactivation regulates membrane excitability by limiting the pool of “activatable” Na_V channels and for fast inactivation is mediated by a group of hydrophobic residues (isoleucine, phenylalanine, and methionine; the IFM motif) localized to the intracellular DIII–DIV linker, forming the “inactivation gate.” In the channel open state, an intracellular binding site for the IFM motif of the inactivation gate is exposed. The subsequent movement and binding of the IFM motif inactivates the channel, as the bound IMF motif blocks the pore of the channel and precludes its reopening until resting state is restored (reviewed in Ref. 255). The mechanisms underlying slow and ultraslow inactivation have not yet been completely described but are thought to involve conformational changes in the pore of the channel (270).

5.1. Persistent Currents Generated by Na_V Channels

In addition to generating the fast-inactivating Na⁺ current (I_NaT) responsible for the rising phase of the action potential (271), Na_V channels are also mediate a “persistent” inward current (I_NaP) essential for sustaining repetitive firing in neurons (214, 272). The Na_V isoforms responsible for I_NaP in all cell types have not been identified. However, Na_V1.6 is a likely contributor in cerebellar Purkinje neurons and in cortical pyramidal neurons (273, 274). In addition to Na_V1.6, the other four Na_V channel isoforms expressed in central neurons all can produce I_NaP.

The molecular mechanism leading to I_NaP is likely to be late channel opening and reopenings during prolonged depolarization, as inferred from the analysis of Na_V1.6 channel currents in heterologous HEK-293 cells (275). To date, however, a detailed investigation of the mechanisms underlying I_NaP in all Na_V isoforms has not been performed. Similarly, there is also a lack of mechanistic information on how native Na_V channel in neurons generate I_NaP. It is possible that I_NaP is facilitated by the cooperative gating of adjacent Na_V channels within homonymous channel clusters, but experimental evidence in support of this hypothesis has not been reported.

5.2. Modulation of Na_V Channel Activity

The majority of the studies investigating the mechanisms modulating Na_V channel activation have focused on the inactivation gate (IG) of the channel. For example, G protein signal transduction pathways activated by monoamines are key regulators of Na_V gating. Physical interaction of G protein βγ subunits with the COOH terminal of the channel have been shown to increase I_NaP (276). Interestingly, Gβγ subunits do not actually alter the voltage dependencies of the rapidly inactivating Na current (I_NaT). Instead, Gβγ subunits shift the voltage dependence of inactivation of I_NaP positively with respect to that of I_NaT. Ma et al. (276) proposed that the Gβγ induces I_NaP by altering fast inactivation of the channel. The mechanism by which this happens seems to involve the formation of Ca²⁺-calmodulin (CaM) bridges from the COOH-terminal IQ motif to the DIII–IV linker. This destabilizes the inactivation gate of the channel, shifting inactivation toward more depolarized potentials (258, 277), and thus increasing the number of noninactivating sodium channels.

A recent study involving the cardiac Na_V1.5 isoform provides an additional potential mechanism for the induction of I_NaP (278). In this model, CaM binds to two independent sites located in the intracellular loop that form the inactivation gate in a Ca²⁺-dependent manner that facilitates recovery from inactivation and also slows down the kinetics of inactivation (278). Future studies should investigate whether a similar mechanism contributes to persistent Na_V channel currents by other isoforms in neurons.

5.3. Expression of Na_V Channels Within Different Excitable Tissues

The primary locations and clustering properties of the different Na_V isoforms have been described in detail in a recent comprehensive review article (279). Our focus here is to identify the cellular compartments of individual cell types where dense aggregations of Na_V channels have been observed and the effects these aggregations may have on cell function. Na_V1.2 and Na_V1.6, for example, are found ubiquitously in neural axons, but Na_V1.6 exhibits a particularly dense concentration at two identified sites: 1) the axon initial segment (AIS), where the high density is presumed to be instrumental in action potential initiation (280) and may contribute to the amplification of synaptic inputs (281), and 2) at nodes of Ranvier in myelinated axons, where the high density supports action potential propagation by saltatory conduction (282). The accumulation and stabilization of Na_V channels at these sites depend on anchorage to the submembrane cytoskeleton through ankyrin G and on adjacent membrane barriers and, in addition, at nodes of Ranvier, axonal cell adhesion molecules that interact with glial membranes to localize and stabilize the node (279). Other Na_V isoforms aggregate in different sites including neuronal dendritic trees, synapses, unmyelinated axons, cardiomyocytes, and other nonneuronal excitable cells (279).

Na_V1.4 channels aggregate on the postsynaptic membrane of neuromuscular junctions of skeletal muscle cells, but the scaffolding and anchoring mechanism remain unresolved. Agrin, a neuronal proteoglycan, appears to mediate the aggregation, at least in cultured rat muscle fibers (283). It has also been suggested that these Na_V1.4 channels are connected to muscle cell cytoskeletal structures via ankyrin-spectrin and to its extracellular matrix through syntrophin-dystrophin-dystroglycan-agrin (284).

Na_V1.5 channels are concentrated in two distinct compartments in cardiomyocytes, the intercalated disk, where Na_V1.5 associates with multiple proteins, including ankyrin G and plakophilin 2, as well as the lateral membranes. Na_V1.5 engages with the syntrophin-dystrophin complex (285). In the lateral membranes of ventricular myocytes, Na_V1.5 channels form clusters of various sizes and densities, the largest of which occur in the crest region of the sarcomere and between which there appears to be little electrical activity (286). With proximity ligation assay (PLA) and stochastic optical reconstruction microscopy (STORM) superresolution microscopy, it has recently been demonstrated that Na_V1.6 channels are not only distributed in dense clusters but colocalize with RyR2 and display noticeable alignment with t-tubules (287). Machine learning-based cluster detection and STORM-relative localization analysis (STORM-RLA) demonstrated that ∼60% of Na_V1.6 clusters are located within 100 nm of RyR2 (288). These results strongly suggest that Na_V1.6 channels are implicated as a component of the Ca²⁺ cycling nanodomain.

Clusters of Na_V1.1 channels associated with ankyrin G and neurofascin have been reported on cells of the inner plexiform layer (IPL) of the retina that are displaced by Na_V1.2 channels during development (289). These clusters in the IPL appear to be restricted to AIS-like processes on AII amacrine cells (290) and likely function as spike generators similarly to the conventional AIS on neurons with full axons.

Finally, in the cochlea, Na_V1.6 channels are clustered on afferent fibers, whereas Na_V1.2 channels are clustered within the membranes of efferent fibers (291). The Na_V1.6 channel clusters are concentrated near the axonal terminals of the primary afferent fibers that contact the inner (type I) and outer (type II) hair cells. In addition, Na_V1.6 is concentrated in the AIS emanating from the type II ganglion cell soma so that, in effect, both the central and peripheral axon processes of these unmyelinated neurons are richly endowed with Na_V1.6 (291). On the efferent cochlea neurons, Na_V1.2 clusters are restricted to the axonal endings.

5.4. Cooperative Gating of Na_V Channels Within Clusters

Beyond identifying where clusters of Na_V channels are found in excitable tissues, we are principally focused in this review on whether Na_V channels within dense clusters form functional oligomeric complexes. Evidence for cooperative gating of neuronal Na_V isoforms, however, is largely confined to modeling studies that suggest that cooperative gating of Na_V channels within dense clusters enhances the capacity of neurons to generate trains of action potentials at high frequencies (292, 293) as well as persistent neural discharge that outlasts the stimulus that triggers it (27).

To our knowledge, the first evidence that this may in fact be the case appeared in a study by Undrovinas and colleagues (9) demonstrating the possible simultaneous opening and closing of two and three channels, presumably Na_V1.5, recorded from cell-attached and excised inside-out patches of rat and rabbit cardiomyocytes. The authors offered no mechanism that might account for the synchronized gating but speculated that this behavior might result from a very tight cooperativity in the gating of separate channels. However, the authors were unable to distinguish this possibility from the alternative hypothesis that these recordings represented supraconductance levels that are only coincidentally the same as integral multiples of the unitary conductance.

It has recently been reported that the α-subunits of Na_V1.1, Na_V1.2, and Na_V1.5 channels are expressed in the membrane as dimers that gate in a cooperative fashion (FIGURE 7, B and C) (10). The two α-subunits interact both directly and indirectly via the cytoplasmic protein 14-3-3. Both mutations at the 14-3-3 interaction sites and inhibition by difopein disengage the two Na_V α-subunits, which has adduced evidence for dimerization and functional interaction for Na_V1.7 channels. As a consequence of these findings, future studies will need to include new analyses and modeling of Na⁺ currents to acknowledge that the quantal unit is generally a dimer rather than a single channel [i.e., the measured Na_V channel conductance is twice the value for an actual single α-subunit pore (294)].

5.5. Pathological Consequences of Altered Clustering of Na_V Channels

Alterations in the normal patterns of Na_V channel distributions are associated with several pathologies. For example, in chronic demyelinating diseases, large clusters of Na_V channels often appear on demyelinated axons (295). Moreover, isoforms other than the normal Na_V1.6, such as Na_V1.7, are often abnormally expressed along the demyelinated axons (296). Similarly, Na_V1.8 channels, which are normally uniformly distributed along the axons of the unmyelinated C fibers innervating nociceptors, display abnormal clustering after injury that appears to be associated with neuropathic pain (297).

In the heart, mutations in SCN5A, the gene encoding cardiac voltage-gated Na_V1.5 channel, lead to a wide variety of cardiac arrhythmias, including long QT syndrome type 3 (LQT3), Brugada syndrome, conduction defects, and sick sinus syndrome. It some cases, the pathology may be caused by an alteration in the distribution of these channels within the lateral membranes and t-tubules (298). In other instances, however, it has been proposed that the mutations in SCN5A may lead to reduced I_Na through a dominant-negative effect exerted by the mutant, nonconducting channels coupling to wild-type channels.

Recent work from the Deschênes laboratory (299) suggests that the heart failure-associated splice variant Na_V1.5-G1642X suppresses Na⁺ currents in heart failure patients by preventing the dimers from gating cooperatively. Similarly, Clatot et al. (300) found that the Na_V1.5-L325R mutant, which is associated with Brugada syndrome, dimerizes with wild-type channels and thus leads to a dominant-negative effect, as reflected by a nearly 75% reduction in Na_V current density. This was the first evidence that cooperative gating of voltage-gated Na⁺ channels can also be responsible for the dominant-negative effect leading to arrhythmias.

Finally, Rühlmann et al. (301) discovered that a mutation in the hNa_V1.7 channel (A1632E) that causes symptoms of erythromelalgia and paroxysmal extreme pain disorder increases persistent Na⁺ currents by dimerizing. Preventing dimerization of hNa_V1.7-A1632E functionally decoupled the channels and decreased persistent Na⁺ currents. Functional uncoupling of mutant hNa_V1.7/A1632E channel dimers restored their defective allosteric fast inactivation mechanism. Thus, as with Ca_V1.2 channels, dimerization of mutant Na_V1/5/1.7 channels could induce or exacerbate pathological conditions.

6. K⁺ CHANNELS

Potassium (K⁺) channels are the most varied and widely distributed ion channels found in the membranes of living cells (2). They are subdivided into four families based on both their structural and functional properties: 1) calcium- or sodium-activated potassium channels (K_Ca or K_Na), 2) inward-rectifying potassium channels (K_Ir), 3) two-pore domain potassium channels (K_2P), and 4) voltage-gated potassium (K_V) channels. More than 80 mammalian genes are engaged in encoding their α protein subunits (302–305).

In contrast to the single polypeptide that forms the ionophore of Na_V and Ca_V channels, all K⁺ channels have a tetrameric structure resulting from the assembly of four identical subunits forming a symmetric (C₄) complex organized to form a conducting pore (i.e., a heterotetramer). In some instances, four related, nonidentical protein subunits assemble to form heterotetrameric channels. All K⁺ channel subunits have a unique pore-loop structure that lines the top of the pore and is responsible for K⁺ selectivity and permeability (306). The principal physiological functions of K⁺ channels include setting the resting membrane potential (V_m), repolarization of excitable cells following the rising phase of an action potential, regulation of the cardiac action potential duration, maintenance of vascular tone, and the regulation of many cellular processes including the secretion of hormones (2).

As is the case for other classes of transmembrane channels, K⁺ channels generally display striking variations in their distributions and relative densities in different cellular compartments and are known to form dense clusters (307, 308). In addition, there have been several reports of alterations in normal gating behavior of K⁺ channels within clusters that we describe in sects. 6.1–6.7 (181, 309–311).

6.1. K_V1 Channels

K_V1 channels are concentrated in the axons and synaptic terminals of neurons (312), where they facilitate action potential propagation (e.g., Ref. 313) and neurotransmitter release (e.g., Ref. 314), respectively. K_V1 channels are also present in the somatodendritic regions of some CNS neurons (315).

At synaptic terminals, K_V1 channels are arranged in distinct clusters that are thought to be formed and maintained through interactions of PSD (postsynaptic density)-95 with a PDZ (postsynaptic density-Dig1-Z0-1) binding motif [X(S/T)XV-COOH] at the channel’s COOH terminus (48, 316). In heterologous cells, the binding induces oligomerization that reduces internalization of expressed K_V1.4 channels and causes them to form clusters (317). Similarly, K_V1.2 channels aggregate with PSD-95 in presynaptic terminals in the cerebellum (48) and are essential for embedding K_V1.4 in rat cortical axons (318). Although PSD-95 is found colocalized with K_V1 channels at juxtaparanodal regions along myelinated axons, it is not essential to generate K_V1 clusters, as they still appeared at the juxtaparanodal regions of the optic nerve in mice that expressed a truncated, mutant form of PSD-95 (319). K_V1.2 clusters are also observed in these mutant mice. At present, it is unclear whether clustering alters K_V1 function.

6.2. K_V2 Channels

K_V2.1 channels are responsible for the delayed-rectifier current in the somatodendritic membranes of neurons (320, 321). They are organized into dense clusters through a phosphorylation-dependent process (322). Furthermore, in hippocampal neurons, it appears that a 28-amino acid sequence on the COOH terminus of the K_V2.1 channels provides the requisite signal and possible template for cluster formation (323).

Recent work on the distribution of K_V2.1 channels has revealed a more complex pattern of functional segregation: half of the channels appear to be broadly distributed within plasma membrane, whereas the other half are found in dense clusters on the soma, dendrites, and axon initial segment (AIS). The clustering occurs at endoplasmic reticulum (ER)-plasma membrane (PM) junctions created via interaction with VAMP-associated proteins in the expressed in the ER (61, 324). What is most intriguing about K_V2.1 channels is that most of these channels are nonconductive in vascular smooth muscle (157) and heterologous expression systems (325). Interestingly, O’Connell, Loftus, and Tamkun (325) proposed that clustered channels are largely nonconducting. This has been referred to as “negative cooperativity” (181, 326). It is unclear, however, whether clustering per se causes negative cooperativity or a parallel signaling event.

6.3. K_V7 Channels

K_V7 channels are slowly activated delayed-rectifier channels that open at subthreshold levels, contributing to the resting membrane potential and reducing excitability (2). K_V7.1 is expressed in multiple cell types such as ventricular myocytes and epithelial cells. K_V7.2 is localized to the axon initial segment (AIS) and nodes of Ranvier in both peripheral and central neurons and is often colocalized with K_V7.3 (327).

A recent study has shown that K_V7.4 channels are arranged in dense clusters in cochlear outer hair cells, within which the channels gate cooperatively (328). The consequences of this cooperative gating are increased gating kinetics and voltage sensitivity. The clustering also appeared to enhance the mechanical sensitivity of the K_V7.4 channels, which varied systematically with hair cell position along the basilar membrane. Cooperative channel gating leading to enhanced electromechanical sensitivity could be induced in HEK-293 cells by fusing channels into clusters by optogenetic methods.

6.4. K_Ca Channels

Large-conductance K_Ca1.1 channels (also known as BK channels) are activated by Ca²⁺ and membrane depolarization (2) These channels have been suggested to assemble into macromolecular complexes with the voltage-gated Ca_V1.2, Ca_V2.1, and Ca_V2.2 channels and couple functionally via local Ca²⁺ signals (329). A recent study suggested that Ca_V1.3 channels could also activate K_Ca1.1 (41). Vivas et al. (41) found that when K_Ca1.1 and Ca_V1.3 channels were coexpressed in the same cell, these channels were expressed into dense clusters of BK channels surrounded by clusters of Ca_V1.3 channels, forming a presumed multichannel complex (41, 329).

Similar patterns of coclustering were observed in a heterologous system as well as in both hippocampal and sympathetic neurons in the rat (39). Although this study did not include electrophysiological measurements, the authors suggest that this is a functional consortium, ensuring “tight tracking between local BK channel activation and the gating of Ca_V1.3 channels at quite negative membrane potentials, facilitating the regulation of neuronal excitability at voltages close to the threshold to fire action potentials.” Future studies should investigate whether, like Ca_V1.3 channels, Ca_V1.2, Ca_V2.1, and Ca_V2.2 clusters form around or within BK channel clusters in neurons.

6.5. Pathological Consequences of Altered Clustering of K_Na1.1 Channels

Mutations of the K_Na1.1 (KCNT1) channels are associated with epilepsy (330). In a recent study (331), it was found that nine different mutant K_Na1.1 channels linked to epilepsy produced many-fold increase in the current produced by wild-type channels. Although wild-type K_Na1.1 channels can gate cooperatively, the authors found that the extent of cooperative gating was much greater for the mutant channels even though their single-channel conductances were reduced (331). These results indicate that changes in the extent to which membrane channels interact in homogeneous clusters may exert as profound an effect on function as changes in the behavior of the individual protein molecules.

6.6. K_2P Channels

K_2P2.1 and K_2P4.1, the thermosensitive and mechanosensitive two-pore potassium channels, have been found to be clustered at the nodes of Ranvier of large-diameter rat trigeminal afferents with density >3,000 times higher than that on the somata (332, 333). These K_2P channels are required to generate the repolarization phase of the propagated action potential at the nodes.

6.7. KcsA Channels

KcsA is a prokaryotic K⁺ channel from the soil bacterium Streptomyces lividans. KcsA is pH activated and has two transmembrane segments that together create a highly selective ion pore (334). Because the amino acid sequence of the selectivity filter of KcsA is highly conserved among both prokaryotic and eukaryotic K⁺ channels, analysis of the structure of this channel has generated important structural and mechanistic insights on the gating mechanisms of K⁺ channels in general. Clusters of KcsA channels have been observed (335), and a recent modeling and experimental study (309) suggests that coupled gating within these channel clusters may account for the unusual electrical activity reported for these channels (334, 336).

7. RYANODINE RECEPTORS

7.1. Structure and Function of Ryanodine Receptors

Ryanodine receptors (RyRs) are Ca²⁺-activated channels expressed in the endo/sarcoplasmic reticulum of many cell types, including neurons and muscle, where they form homotetrameric assemblies (82, 86, 154). The molecular mass of a RyR tetramer is ∼2.2 MDa. The RyR channel has two distinctive domains (337–339). The cytoplasmic domain comprises the largest portion of the channels (∼29 × 29 × 12 nm). Multiple proteins bind to this cytosolic domain (e.g., calmodulin and FKBP12). The second domain is relatively small (∼7 nm) and includes the channel’s transmembrane segments where the pore is located (FIGURE 8A). A detailed review of RyR structure and function has recently been published (340). Here, we focus our discussion on clustering and cooperative gating of RyRs.

FIGURE 8.

Ryanodine receptor (RyR) structure, organization, and a model of RyR interactions with other ion channels. A: illustration of an RyR subunit and RyR surface map highlighting its 3-dimensional structure. RyRs organize into clusters or arrays that enable cooperative gating. Surface map and structure based on Gong et al. (449) and organization based on Yin et al. (450). CaM, calmodulin; SR, sarcoplasmic reticulum. B: model of RyR organization in wild-type (WT) and mdx cells and its impact on large-conductance (BK) channel activity in vascular smooth muscle. STOCs, spontaneous transient outward currents. C: model of the organization and communication between RyR and Ca_V1.2 in the t-tubules of cardiac myocytes. Figure created with Biorender.com.

There are three isoforms of RyRs: types 1 (RyR1), 2 (RyR2), and 3 (RyR3). They differ in their tissue and cellular compartment distributions. RyR1s are expressed in skeletal muscle, RyR2s are expressed predominantly in cardiac and smooth muscle, and RyR3s are expressed in neurons as well as smooth muscle (341–343). RyR3 is also expressed at low levels in striated muscle. In general, the shapes and binding sites of modulatory proteins (e.g., calmodulin, FK506-binding proteins) of RyR channels are similar (344, 345).

As intracellular channels, the biophysical properties of RyRs are difficult to study with standard patch-clamp electrophysiological approaches. Accordingly, the most detailed biophysical analyses of RyRs have been done by incorporating them into planar lipid bilayers (346, 347). The three RyR isoforms have many biophysical properties in common, including activation by submicromolar Ca²⁺ elevations in the cytosolic side of the channel, by ATP, and by caffeine, as well as inhibition by micromolar Mg²⁺ concentration, ruthenium red, and procaine (341–343). RyRs also display comparable unitary conductances under similar experimental conditions. However, there are some important differences between isoforms, such as the higher sensitivity of RyR2 and RyR3 to cytosolic Ca²⁺ than RyR1. RyR3s are also more resistant to inactivation by high Ca²⁺ concentrations (348).

Our understanding of the mechanisms controlling RyR function in intact cells took a major step forward following the development of bright, high-dynamic range Ca²⁺ indicators such as fluo-3. Cheng et al. (127) used confocal microscopy with high, diffraction-limited resolution to image intracellular Ca²⁺ in intact rat ventricular myocytes loaded with fluo-3. They discovered that these cells produced localized intracellular Ca²⁺ elevations, which they called Ca²⁺ sparks, consequent to the opening of RyR2s (reviewed in Ref. 349). A key feature of Ca²⁺ sparks was that their amplitude distribution was unimodal, suggesting they were quantal Ca²⁺ release events. The Ca²⁺ flux associated with a spark was ∼3 pA, similar to the Ca²⁺ current produced by the opening of a single RyR in a planar lipid bilayer (350–352). This similarity led them hypothesize that Ca²⁺ sparks could be produced by the opening of a single RyR or possibly a small cluster of RyRs acting in concert and that Ca²⁺ sparks are the elementary events underlying excitation-contraction coupling in the heart. Accordingly, a local control model for excitation-contraction coupling was proposed in which the amplitude of whole cell [Ca²⁺]_i transient is graded by the number of Ca²⁺ sparks activated by Ca²⁺ influx via L-type Ca_V1.2 channels (127, 142, 143). Following the publication of the seminal Cheng et al. study (127), Ca²⁺ sparks have been imaged in multiple cell types, including skeletal muscle (355, 356), smooth muscle (241, 357), and neurons (49, 358).

7.2. Cooperativity Among RyRs

Although the hypothesis that Ca²⁺ sparks are the elementary events underlying cardiac excitation-contraction coupling was broadly accepted, the issue of whether these events are produced by the opening of a single RyR or a small cluster of RyRs opening simultaneously was the focus of multiple follow-up papers. In an elegant set of experiments, Michael Fill and his team (359) recorded elementary RyR currents in planar lipid bilayers but, unlike in prior studies, under near-physiological conditions (e.g., physiological Mg²⁺ concentration and Ca²⁺ gradient, ATP). Their data indicated that the RyR single-channel current was ∼0.25 pA, an order of magnitude smaller than the currents recorded under nonphysiological conditions (∼3 pA). These data strongly suggested that normal Ca²⁺ sparks were likely produced by a cluster of RyRs opening in synchrony.

The results described above were consistent with classic electron microscopy studies (87, 360, 361) and, more recently, fluorescence superresolution microscopy (28, 49, 133, 362) showing that RyRs organize in clusters/arrays in the endo/sarcoplasmic reticulum of muscle and neurons (FIGURE 8A). In cardiac and smooth muscle, it has been estimated that these clusters contain ∼20–40 ryanodine receptors, respectively (133, 362, 363). Because RyR clusters in neurons subsume about the same area as those observed in muscle cells, the expectation that a similar number of channels is contained in these clusters is quite reasonable. Importantly, the current models of junctional sarcoplasmic reticulum suggest that Ca²⁺ sparks may be the result of the concerted opening of multiple RyR clusters that collectively form a functional “supercluster” (28, 133).

As Ca²⁺-activated channels, a plausible mechanism by which a cluster of RyRs could be activated at or near synchrony is Ca²⁺-induced Ca²⁺ release. One possibility is that an entire RyR cluster is activated by an adjacent Ca_V1.2 channel cluster. Similarly, Ca_V1.2 channels could activate one or a few RyRs in the jSR, which then subsequently activate neighboring RyRs via the Ca²⁺-induced Ca²⁺ release mechanism. In this model, Ca²⁺ is the coupling signal for clustered RyRs. Indeed, Porta et al. (364) found that RyR1 clusters could also undergo cooperative gating in planar lipid bilayers and that the coupling strength was differentially modulated by cytosolic ATP, Mg²⁺, and Ca²⁺. Extensive in silico work has been done to simulate how ryanodine-to-ryanodine receptor coupling might operate (365–367).

Marx et al. (88, 368) offered an alternative model of RyR coupling. They showed that multiple RyR1s and RyR2s can be isolated under conditions such that they remained physically coupled to one another in planar lipid bilayers and coordinated their channel openings. The key finding is that the regulatory subunit FK506 binding protein (FKBP12 in skeletal muscle and FKBP12.6 in cardiac muscle) functionally couples multiple RyR1/2 channels. Although the simultaneous activation of RyRs could be mediated entirely through activation of clustered channels by locally released Ca²⁺, it has been proposed that physical allosteric interactions can also contribute to RyR gating and thus Ca²⁺ sparks. This FKBP12-based coupling model was extended to suggest that loss of FKBP12/FKBP12.6-decoupled RyRs and increased sarcoplasmic reticulum Ca²⁺ leak contributes to contraction failure and arrhythmias in the heart (369, 370) and fatigue in skeletal muscle (371). Interestingly, detachment of FKBP12.6 from the RyR2 channel does not lead to the elimination of Ca²⁺ sparks. Instead, it increases Ca²⁺ spark frequency and duration (372). Although the bilayer data supporting physical coupling of RyRs are quite compelling, the exact mechanisms by which the opening of RyR channels within clusters are coordinated remain poorly understood.

An important observation made by Baddeley et al. (28), Soeller et al. (133) and Pritchard et al. (363) is that RyRs cluster sizes follow a near-exponential distribution compatible with a stochastic self-assembly process for channel cluster formation similar to that of Ca_V1.2 and large-conductance Ca²⁺-activated K⁺ channels (5). This is significant because it suggests that the two key components of the dyad in cardiac and smooth muscle are formed randomly. Di Maio et al. (373) proposed a model for the formation of Ca²⁺ release units in striated muscle in which the first step involves the emergence of t-tubules. Stable sarcolemma-sarcoplasmic reticulum junctions are formed by junctophilin-2 (363, 374, 375). This is followed by the recruitment of the junctional SR proteins such as triadin, RyRs, and Ca_V1.2/1.1 channels but could lead to “silent” dyads if these proteins are not delivered to these sites.

Recent studies on t-tubules considered in conjunction with those on the formation of Ca_V1.2 and RyR channel clusters suggest an alternative model. t-Tubule formation and random Ca_V1.2 clustering are possibly early events in the formation of functional Ca²⁺ release units. Several proteins have been identified to play important roles in t-tubule and dyad formation. These include Caveolin-3 (376–378), dysferlin (379, 380), junctophilin 2 (381), triadin (382), and BIN1 (383–385). BIN1 is a member of the Bin1-Amphiphysin-Rvs (BAR) domain superfamily BIN1, which has been shown to promote t-tubule biogenesis and to serve as anchoring point for microtubules in cardiac muscle. In ventricular myocytes, BIN1 is critical for the trafficking of Ca_V1.2 channels to the sarcolemma from the Golgi apparatus (383–386). This not only allows for the delivery of Ca_V1.2 channels to the t-tubules but may also direct the movement of jSR along microtubules anchored to BIN1 toward newly formed Ca_V1.2 clusters (387). Thus, although RyR-expressing junctional SR membrane may randomly contact the sarcolemma, it only results in stable dyads where Ca_V1.2 (e.g., cardiac) or BK channel (e.g., smooth muscle) clusters are randomly formed.

A similar process may be at play in ER-plasma membrane junctions in neurons. K_V2.1 channels promote the clustering of Ca_V1.2 channels in the somata of hippocampal neurons. Furthermore, RyR-expressing ER terminals exist at sites where K_V2.1 and Ca_V1.2 channels cocluster (49). It is possible that ER movement is stochastic (388), but, as in muscle, stable ER-plasma membrane junctions only form where K_V2.1 and Ca_V1.2 channels cocluster. The key concept here is that multiple stochastic processes (i.e., Ca_V1.2 clustering, BK clustering, RyR clustering, and SR/ER mobility) may lead to dyad formation in nerve and muscle.

7.3. Physiological Implications of RyR Cooperative Gating

A central tenet of intracellular Ca²⁺ signaling is that RyR and IP₃R recruitment is hierarchical as follows. Ca²⁺ sparks are produced by the simultaneous activation of clustered RyRs. Each of the RyRs within these clusters produces an elementary Ca²⁺ signal, called a “Ca²⁺ quark” (389, 390). Thus, Ca²⁺ quarks produce Ca²⁺ sparks. Whole cell Ca²⁺ transients are produced by the temporal and spatial summation of multiple independent Ca²⁺ spark sites (i.e., Ca²⁺ release units). RyR-mediated Ca²⁺ waves are produced and propagated by the successive, progressive activation of Ca²⁺ spark sites along the cell. Thus, current Ca²⁺ signaling models are based on this Ca²⁺ quark, Ca²⁺ spark, whole cell Ca²⁺ transient/Ca²⁺ wave hierarchy.

Alterations in the behavior of any of the components of this hierarchical signal cascade could have important implications in Ca²⁺-dependent cellular events. For example, an increase in the coupling strength between RyRs within a cluster could increase the open probability of these channels and hence increase the amplitude and spatial profile of Ca²⁺ sparks. This could result in larger Ca²⁺ transient in cardiac myocytes and hence contraction during the action potential. An increase in Ca²⁺ spark amplitude could increase the probability of activating nearby Ca²⁺ release units via Ca²⁺-induced Ca²⁺ release (CICR), which could result in a Ca²⁺ wave. An increase in Ca²⁺ spark amplitude could increase the probability of Ca_V1.2 channel coupling, Ca_V1.2 Ca²⁺-dependent inactivation, in cardiac myocytes or BK channel activation in smooth muscle. An increase in RyR cluster size would have a similar impact on Ca²⁺ signaling even in the absence of an increase in coupling strength.

As an example, Pritchard et al. (363) found that RyR2 and BK channels were organized into clusters. The RyR2 clusters that colocalized with BK channel clusters were nearly 1.6-fold larger in smooth muscle cells from the mouse Duchenne muscular dystrophy model (mdx) than in control myocytes (FIGURE 8B). They found that the amplitude and frequency of Ca²⁺ sparks were greater in smooth muscle cells from mdx mice than those from control mice. This finding was associated with a near loss of myogenic tone in cerebral pial arteries, and parenchymal arterioles from mdx mice failed. However, inhibition of RyR and BK channel activity increased myogenic tone in mdx arteries to control levels. Thus, increased size of RyR2 protein clusters increases Ca²⁺ spark and BK channel activity and could lead to cerebral microvascular dysfunction.

Similarly, Xie et al. (367) found that the Ca²⁺ spark frequency and amplitude increased in a nonlinear fashion as the size of RyR clusters increased in ventricular myocytes (FIGURE 8C). The larger RyR clusters are activated at lower Ca²⁺ levels. However, the larger fluxes associated with larger RyR clusters led to lower SR Ca²⁺ content. By contrast, the smaller RyR clusters had a higher threshold for activation and a lower associated leak. This increased SR Ca²⁺ content.

Mathematical modeling suggested that homogeneously large or small RyR clusters decrease the propagation of Ca²⁺ waves due to low load for large clusters but relatively low excitability for smaller RyR clusters (367). By contrast, their model predicted that expression of a combination of large and small RyR clusters would result in the potentiation of Ca²⁺ waves, as the enhanced SR Ca²⁺ load caused by smaller RyR clusters enables Ca²⁺ wave initiation and propagation from larger RyR clusters. Thus, a heterogeneous distribution of RyR cluster size may increase Ca²⁺ wave production and propagation and thus afterdepolarizations and triggered arrhythmias.

Kolstad et al. (391) investigated whether the deficit described above could be traced to nanoscale RyR reorganization. They found that failing myocytes had more numerous, smaller RyR clusters compared with those extracted from control rats. The decrease in the size of Ca²⁺ release units was associated with an increase in optically “silent” SR Ca²⁺ leak and a decrease in Ca²⁺ spark numbers during heart failure. Ca²⁺ sparks in failing cells displayed slow kinetics, as the Ca²⁺ diffused across dispersed Ca²⁺ release units. These slow Ca²⁺ sparks desynchronized the whole cell Ca²⁺ transient during the action potential. Thus, Kolstad et al. (391) elegantly demonstrated that reorganization of RyRs increases Ca²⁺ leak and contributes to lower contractility during the development of heart failure.

8. IP₃ RECEPTORS

The inositol trisphosphate (IP₃)/Ca²⁺ signaling pathway regulates many cellular processes, including metabolism, secretion, fertilization, proliferation, smooth muscle contraction, memory, and insulin secretion from pancreatic beta cells. Berridge (392), Foskett et al. (393), Narayanan et al. (394), and Woll and Van Petegem (340) have written excellent and extensive articles on inositol 1,4,5-trisphosphate/calcium (IP₃/Ca²⁺) signaling and IP₃ receptor (IP₃R) structure, function, and pathology in different cell types. Here, we focus our discussion on clustering and cooperative gating of IP₃Rs.

8.1. IP₃R Structure and Function

Like RyRs, IP₃Rs are large tetrameric intracellular proteins. Each IP₃R subunit has an apparent molecular mass of ∼300 kDa (395). The cytoplasmic NH₂ terminus of each InsP₃R molecule accounts for nearly 85% of the total protein mass. The IP₃Rs contain six membrane-spanning helices that form the pore of the channel (FIGURE 9A). They also have a relatively short cytoplasmic carboxy terminus. The NH₂-terminal region is where the proximal IP₃ binding domain is located (396). Three IP₃R isoforms have been identified: IP₃R types 1–3 (IP₃R1, IP₃R2, and IP₃R3). All three IP₃R isoforms cluster and have similar conductances and similar gating kinetics in intact cells (397).

FIGURE 9.

Topology of inositol trisphosphate (IP₃) receptors (IP₃Rs) and models of IP₃R interactions with other ion channels. A: structure of an IP₃R subunit. SR, sarcoplasmic reticulum. B: model. IP₃ (1) promotes (2) a physical interaction between IP₃Rs and canonical transient receptor potential (TRP) (TPRC)3 channels that (3) activates TRPC3 and promotes membrane depolarization, resulting in (4) an increase in the open probability of Ca_V1.2 channels, an elevation in intracellular Ca²⁺, and contraction of vascular smooth muscle cells. C: stretch promotes IP₃ production and activation of IP₃Rs, as well as diacylglycerol (DAG) production leading to activation of TRPC6 channels. The activation of IP₃Rs and TRPC6 channels increases local intracellular Ca²⁺ near melastatin TRP (TRPM)4 channels, which are then activated to cause membrane depolarization and vascular smooth muscle contraction. In this model, it is proposed that IP₃Rs, TPRC6, and TRPM4 form a nanocomplex. Figures created with Biorender.com.

Neurotransmitters, vasoactive peptides, and growth factors induce the synthesis of IP₃ through the activation of G_q protein-coupled receptors (G_qPCRs) or the protein tyrosine kinase-linked receptors that are linked to different phospholipase C (PLC) isoforms (reviewed in Ref. 398). Activation of PLC hydrolyzes the lipid phosphatidylinositol 4,5-bisphosphate, producing diacylglycerol and IP₃. IP₃ binds and thereby opens IP₃Rs to release Ca²⁺ from the endo/sarcoplasmic reticulum.

An interesting characteristic of IP₃Rs is that they are dually activated by IP₃ and Ca²⁺. However, the efficacy of Ca²⁺ activation of IP₃Rs depends on IP₃ levels. The cytoplasm has been referred to as an “excitable medium” in which IP₃R activation is modulated by IP₃ levels and Ca²⁺-induced Ca²⁺ release (CICR) (399). IP₃Rs, like RyRs, are a critical component of the excitable medium, and their excitability depends on the proximity between IP₃Rs as well as local IP₃ and Ca²⁺ levels (399–402).

A key feature of the excitable medium model is that it facilitates the rapid propagation and amplification of local Ca²⁺ signals via CICR. As proposed by Foskett et al. (393), however, subcellular variations in distance between release sites and excitability of the system may limit the efficacy of local signals to activate nearby IP₃Rs via CICR and hence prevent all-or-nothing Ca²⁺ signaling events. As recently shown by Longden et al. (403), this type of hierarchical IP₃R-mediated Ca²⁺ signaling cascade has important implications in the regulation of the membrane potential of endothelial cells of brain capillaries and hence arteriolar diameter and blood flow during neural activity.

8.2. Cooperativity Among IP₃ Receptors

As is the case with RyRs, the spatial organization of IP₃Rs and their sensitivity to IP₃ and Ca²⁺ confer a great deal of flexibility in Ca²⁺ signaling regulation. Local interactions between IP₃Rs allow the hierarchical recruitment of IP₃Rs so that single IP₃Rs respond first, producing “Ca²⁺ blips,” and then clustered IP₃Rs can open together, producing a “Ca²⁺ puff.” As Ca²⁺ puffs become more frequent, they give rise to Ca²⁺ waves (402, 404).

Superresolution imaging suggests that IP₃Rs are randomly distributed within the ER, forming clusters of various sizes (56–58). Most IP₃R clusters are mobile, however. Low concentrations of IP₃ cause IP₃Rs to rapidly aggregate reversibly into clusters generally consisting of four closely associated IP₃Rs (56). Dickinson et al. (405) proposed that the stochastic opening of any individual IP₃R channel within a cluster triggers a Ca²⁺ puff by activating neighboring IP₃Rs via CICR.

When intracellular [Ca²⁺] is low (i.e., ∼100 nM), most, if not all, clustered IP₃Rs are predicted to open independently and with a relatively low open probability (56). Clustered IP₃Rs have also shorter open times and appear to be less sensitive to IP₃ than single IP₃Rs. However, as cytosolic Ca²⁺ increases, the gating of clustered IP₃Rs becomes cooperative. Dynamic regulation of clustering by IP₃ retunes IP₃R sensitivity to IP₃ and Ca²⁺. This facilitates the recruitment of the elementary IP₃R Ca²⁺ release events.

Consistent with the model outlined above, Wiltgen et al. (404) found that the kinetics of IP₃R channel closing differs from that expected for independent gating. They found that upon activation clustered IP₃Rs remained open in synchrony for periods much longer than predicted from their individual open lifetimes and that they close simultaneously. Wiltgen et al. (404) proposed that the gating of closely adjacent IP₃Rs could not be explained by stochastic attrition mechanisms, Ca²⁺-inhibition of IP₃Rs, and/or local depletion of Ca²⁺ in the ER lumen. Instead, they postulated that clustered IP₃Rs are coupled, likely via allosteric interactions. At present, however, the mechanisms underlying allosteric regulation of clustered IP₃Rs are remain unresolved.

8.3. Coupling of IP₃ Receptors to Other Channels

IP₃Rs, like RyR1s, appear to physically interact with other sarcolemmal channels. Adebiyi et al. (244) found that IP₃R1s are positioned in close spatial proximity to sarcolemmal TRPC3 channels. Interestingly, activation of phospholipase C by Endothelin-1 increased IP₃ and thus induced direct physical coupling between IP₃R1 and sarcolemmal TRPC3 channels in arterial myocytes. This coupling activated TRPC3 channels, which depolarized arterial myocytes and led to the activation of Ca_V1.2 channels, increased cytosolic Ca²⁺ levels, and thus vasoconstriction (FIGURE 9B).

In a separate study, Gonzales et al. (247) proposed an indirect path for IP₃R-TRPC6/TRPM4 coupling, which is also critical for the development of myogenic tone by arterial smooth muscle in response to increases in intravascular pressure. They demonstrated that an increase in intravascular pressure stretches the sarcolemma of arterial myocytes and activates phospholipase Cγ1. The IP₃ produced by the activation of phospholipase Cγ1 makes the junctional SR IP₃Rs more sensitive to the Ca²⁺ entering the cell via TRPC6 channels. This increases IP₃R-mediated Ca²⁺ release, which activates TRPM4 channels. These TRPM4 currents were found to depolarize smooth muscle cells and cause vasoconstriction (FIGURE 9C).

8.4. Physiological Implications of IP₃ Receptor Cooperative Gating

The coupling of IP₃Rs amplifies and, depending on the coupling strength, grades the amplitude, spatial spread, and frequency of intracellular Ca²⁺ signals. This could result in stronger contractions in vascular smooth muscle. In neurons and nonexcitable cells, the magnitude of IP₃R coupling might alter gene expression and excitability. Notably, a recent study (406) suggested that changes in the nanostructural organization and activity of IP₃R1 is linked to cell death in Niemann-Pick type C1 disease. Future studies should investigate whether changes in the clustering of IP₃Rs contribute to other pathological conditions via changes in cooperative gating and IP₃-mediated Ca²⁺ signaling in different cell types.

9. HYPERPOLARIZATION-CYCLIC NUCLEOTIDE-ACTIVATED CHANNELS

9.1. HCN Channel Structure and Function

Hyperpolarization-cyclic nucleotide-activated (HCN) channels are tetrameric proteins. Each protomer contains six α-helices (S1–S6) in the transmembrane domain, including a voltage sensor domain composed of S1–S4 and a pore domain composed of S5, a reentrant pore region, and S6. Although HCN channels are members of the voltage-gated potassium channel K_V superfamily, they differ from K_V channels in several important ways. Most significantly, HCN channels are opened by hyperpolarization rather than by depolarization (407) and HCN channel opening is facilitated by binding of cyclic nucleotides to a COOH-terminal cytosolic cyclic nucleotide binding domain (408, 409).

To date, four HCN isoforms (HCN1–4) have been identified in humans and mice (410). HCN1 is expressed in the hippocampus, neocortex, and cerebellar cortex but also the heart. HCN2 is expressed in midbrain and thalamic neurons. Loss of HCN2 expression nearly eliminates pacemaker currents in thalamocortical neurons and sinoatrial node (SAN) myocytes (411). HCN3 is expressed in ventricular cells (412). HCN4 is the main isoform expressed in the SAN of the heart as well as the thalamic nuclei and basal ganglia (413–416). In addition, several studies have suggested that HCN channels form by heteromultimers: HCN1/HCN2 form heteromeric channels (417), and both HCN1/HCN4 and HCN2/HCN4 heteromultimers have been detected in the SAN and heterologous expression systems (418, 419).

Single-channel recordings of pacemaking currents in native SAN cells revealed single-channel conductance of ∼1 pS for HCN channels, which is very low compared with other voltage-dependent cation channels (77). This is important because it imposes a technical challenge in recording elementary HCN currents with a high signal-to-noise ratio. For this reason, nonstationary fluctuation analysis has been used to estimate the conductance of cloned HCN channels. With this approach, the conductance of HCN2 was estimated to be ∼2.5 pS (160) and ∼0.7 pS for channels underlying HCN currents in neuronal dendrites (178) Thus, the elementary currents produced by most, if not all, HCN channel isoforms are relatively small. This suggests that to alter membrane potential significantly a relatively large number of HCN channels must open simultaneously and/or that the input resistance of the cell membrane be relatively high.

9.2. Cooperativity Among HCN Channels

Dekker and Yellen (8) performed the first rigorous single-channel study of HCN2 channels. Using a nonstationary fluctuation analysis, they detected discrete HCN2 channel openings with amplitudes ranging from 150 to 230 fA. The mean conductance of these openings was ∼1.5 pS at −120 mV in a symmetrical 160 mM K⁺ bathing solution. A key finding in their study was that multiple HCN2 channels would undergo seemingly correlated openings, suggesting cooperativity between channels.

Dekker and Yellen (8) speculated about the potential mechanisms underlying the coupling of HCN channels. One possibility they raised was that channel coupling resulted from common sensing of a signaling molecule such as cAMP. Experiments performed in excised cell-free patches in both 0 cAMP and saturating 1 mM cAMP, however, ruled out this possibility. Accordingly, Dekker and Yellen (8) proposed that the most likely mechanism for coupling involved direct allosteric communication through direct channel-to-channel interactions mediated by an adaptor or linking protein.

Although similar experiments have not been performed on the HCN1–4 isoforms of the channels, their structural and functional similarities suggest they could also be capable of undergoing cooperative gating. Furthermore, the possibility that HCN1/HCN2 and HCN2/HCN4 could form heteromultimers raises the intriguing possibility that HCN2 may act as coupling agent between HCN1/HCN2 and HCN2/HCN4 heteromultimers (417, 419, 420). Future experiments should determine whether, and under which conditions, native HCN channels undergo cooperative gating in neurons and SAN myocytes.

9.3. Physiological Implications of HCN Cooperative Gating

An important physiological implication of HCN cooperativity is its potential role in modulating the frequency of oscillatory networks in the heart and brain. Here, we focus on the heart’s SAN cells, as they express HCN channels in addition to Ca_V1.2 and Ca_V1.3 channels and ryanodine receptors, all of which may undergo cooperative gating. The principal function of the cells in the SAN of the heart is to produce synchronous action potentials (APs) that initiate each heartbeat. This is accomplished despite the findings that the intrinsic excitability of its myocytes and even its vasculature is highly variable (416, 421). Some cells produce APs fairly frequently; others have periods of discharge separated by long periods of inactivity (i.e., bursts). Some cells only produce stochastic spontaneous subthreshold voltage fluctuations, and a relatively small fraction of cells are electrically silent. The combination of diverse electrical modalities within the SAN cell population leads to highly periodic oscillations through a mechanism of entrainment (421, 422) or stochastic resonance (416, 423).

The initiation of the AP in SAN myocytes occurs during diastole and is driven by activation of HCN channel (424, 425) currents (409, 426, 427). Thus, cooperative gating of clustered HCN channels may increase the amplitude of the funny current (I_f) and hence the rate of diastolic depolarization of spiking pacemaking cells. It is tempting to speculate that the spontaneous subthreshold voltage fluctuations observed by Grainger et al. (416) in SAN myocytes may be produced by the simultaneous opening of coupled HCN channels and to wonder whether or not cooperative gating is more prominent in some SAN cells than others.

Concurrent with HCN activation, Ca²⁺ sparks (127) activate inward diastolic Na⁺/Ca²⁺ exchanger currents (429, 430). These HCN and Na⁺/Ca²⁺ exchanger currents depolarize SAN myocytes beyond the activation threshold of voltage-gated Ca_V3.1 (431), Ca_V1.3 (201, 432), and, eventually, Ca_V1.2 (432, 433) channels. This initiates the AP and triggers a transient global increase in [Ca²⁺]_i. Repolarization of the membrane potential during the AP begins by the inactivation of Ca²⁺ currents and activation of voltage-gated K⁺ currents. As SAN myocytes approach hyperpolarized potentials, HCN channels are activated, initiating this regenerative process.

In summary, the cooperative activating of HCN, Ca_V1.2, Ca_V1.3, and ryanodine receptors influences SA node excitability and hence heart rate. Future experiments should investigate the spatial organization of these channels, whether cooperative gating of these channels varies throughout the node, and how it is influenced by the autonomic nervous system and potentially altered in disease.

10. OTHER CHANNELS REPORTED TO UNDERGO COOPERATIVE GATING

Although we focus on Ca_V, TRP, Na_V, K⁺, and HCN channels as well as RyRs and IP₃Rs in this review, multiple studies have reported nonindependent gating of other channels (see for example Ref. 181). For example, ligand-gated channels including nicotinic acetylcholine receptors (434, 435), NMDA receptors (436), and P_2X receptors (437) have also been shown to gate nonindependently. In the case of acetylcholine receptors, the changes in the oligomeric state correlated with receptor activity and corresponding alterations in postsynaptic currents (438). A recent study suggests that clustering of acetylcholine receptors in the postsynaptic membrane is facilitated by their interaction with the scaffold protein rapsyn (439), thus providing a potential structural arrangement that enables coupling of receptors. More recently, there is debate on whether the mechanosensitive Piezo1 channel may undergo cooperative gating (440, 441). Intriguingly, the application of Yoda 1, which is an allosteric modulator of the channel, increases Piezo1 clustering (442). Whether Piezo1 channels undergo cooperative gating and the physiological implications of its clustering remain to be established.

11. SUMMARY, OPEN QUESTIONS, AND FUTURE DIRECTIONS

In this review, we have presented the evidence for cooperative gating of multiple ion channels, discussed the putative mechanisms underlying the phenomenon, and speculated on the physiological consequences of this gating modality. The preponderance of the data suggest that cooperative gating is not a rare event but rather a consequential mode of gating in a wide range of ion channels. Cooperative gating seems to affect every aspect of electrical and Ca²⁺ signaling in excitable and nonexcitable mammalian cells, i.e., pacemaking, depolarization phase of APs, repolarization, and ER/SR Ca²⁺ release (see FIGURE 10).

FIGURE 10.

Proposed physiological implications of the cooperative gating of ion channels within membrane clusters. AP, action potential; EC, excitation-contraction; ET, excitation-transcription; βAR, β-adrenergic receptor. Figure created with Biorender.com.

To date, a great deal of evidence for cooperative gating has been adduced from the results of biophysical (i.e., patch-clamp electrophysiology, planar lipid bilayers) and imaging (i.e., FRET, bimolecular complementation assays, optogenetic fusion of channels) experiments. At present, however, detailed structural work on how adjacent channels undergo physical, allosteric interactions that coordinate their openings is scarce. Exceptions are Na_V and Ca_V1 channels, where in both cases the COOH terminal of the channel has been identified as critical for channel-to-channel communication. How exactly the physical contact of two or more Ca_V1 or Na_V channels via their C-tails translates into the increased probability of opening the channels gates remains unclear but may involve S4 transitions, as high-P_o channels dominate linked low-P_o channels. It is also conceivable that the Ca•CaM linkages between the COOH termini of adjoined channels result in mechanical forces that are transferred to the channel ionophores to induce channel openings.

We know very little about the structural determinants of TRP, HCN, RyR, and IP₃R channel coupling. We anticipate that emerging crystallographic and cryo-EM structural data from ion channels combined with structural modeling (e.g., Rosetta and/or molecular dynamics) will provide clues as to the chain of events translating physical contact at specific channel domains to channel gate opening. Importantly, these analyses should lead to the formulation of testable predictions on the mechanisms underlying the coupling of specific channel types in living cells.

Another promising area of investigation is the incorporation of cooperative gating into mathematical models of AP generation, propagation, and Ca²⁺ signaling that has already begun in the neurosciences (27, 181, 188, 292, 293, 443, 444). It will be extremely interesting to learn how incorporating cooperative gating of HCN, Ca_V1.3/1.2, and RyRs into mathematical models of cardiac pacemaking, EC coupling, and arrhythmogenesis will inform our understanding of these primary physiological phenomena. In arteries, modeling the impact of cooperative gating of Ca_V1.2, TRP, IP₃R, and RyRs on smooth muscle myogenic tone and propagation of electrical signals through endothelial cells during neurovascular coupling in health and disease may also yield important insights.

To conclude, nearly 70 years after the seminal work by Hodgkin and Huxley established the foundations of modern biophysics with the advent of their quantitative model for action potential generation and propagation in neurons, a growing body of evidence suggests that ion channel gating is not exclusively stochastic and independent. Rather, cooperative gating of ion channels seems to be a general phenomenon that modulates multiple physiological processes including excitability, contraction, arterial diameter, and neuronal coding. Future work on the structural mechanisms together with modeling of the physiological impact of cooperative gating of ion channels in organ function will provide critical insights into how the frequency, duration, and strength of this gating modality is regulated.

GRANTS

This work was supported by grants from the National Institutes of Health to L.F.S. (1R01HL144071, 5R01HL152681, 5R01NS114210, 5R01HL085686, and 1OT2OD026580), R.E.D. (R01AG063796), M.D.B. (1R01NS110953), and M.F.N. (R01HL098200, R01HL121059, R01HL149127, and R01HL161872, and UC-MEXUS CONACYT award).

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the authors.

AUTHOR CONTRIBUTIONS

R.E.D., M.F.N., M.D.B., and L.F.S. interpreted results of experiments; prepared figures; drafted manuscript; edited and revised manuscript; and approved final version of manuscript.