Intercellular Ca²⁺ Waves: Mechanisms and Function

Abstract

Intercellular calcium (Ca²⁺) waves (ICWs) represent the propagation of increases in intracellular Ca²⁺ through a syncytium of cells and appear to be a fundamental mechanism for coordinating multicellular responses. ICWs occur in a wide diversity of cells and have been extensively studied in vitro. More recent studies focus on ICWs in vivo. ICWs are triggered by a variety of stimuli and involve the release of Ca²⁺ from internal stores. The propagation of ICWs predominately involves cell communication with internal messengers moving via gap junctions or extracellular messengers mediating paracrine signaling. ICWs appear to be important in both normal physiology as well as pathophysiological processes in a variety of organs and tissues including brain, liver, retina, cochlea, and vascular tissue. We review here the mechanisms of initiation and propagation of ICWs, the key intra- and extracellular messengers (inositol 1,4,5-trisphosphate and ATP) mediating ICWs, and the proposed physiological functions of ICWs.

I. INTRODUCTION

Intercellular calcium (Ca²⁺) waves (ICWs) consist of increases in cytoplasmic Ca²⁺ ion concentration ([Ca²⁺]_i) that are communicated between cells and appear as waves that spread radially from an initiating or trigger cell (FIGURE 1). In contrast, Ca²⁺ waves that only propagate within a single cell are called intracellular Ca²⁺ waves. The speed and size of ICWs depend on the nature and strength of the initiating stimulus as well as the mechanism of propagation. ICWs often propagate for periods of up to tens of seconds with speeds of ∼10–20 μm/s and consequently can involve tens to hundreds of contiguous cells (53, 96, 122, 123, 175, 306, 312). In some tissues that form an intrinsically excitable cell medium, such as hippocampal slices that consist of a mix of electrically excitable neurons and chemically excitable glia, ICWs can propagate with a spiral path (145).

Figure 1.

Examples of intercellular Ca²⁺ waves (ICWs) in cell cultures. A: an ICW in C6 glioma cells transduced to express the gap junction connexin, connexin (Cx) 26. The ICW was initiated by the focal photolytic release of inositol 1,4,5-trisphosphate (IP₃) within a cell (panel 2s) at the location indicated by the star symbol. B: an ICW in rat brain endothelial cells (endogenously expressing Cx37 and Cx43) induced by exposure of the cells to extracellular Ca²⁺-free conditions. Dimension bars = 50 μm. Color bar indicates change in fluorescence level. An increase in fluorescence represents an increase in [Ca²⁺]_i. Sample time of each panel is indicated in seconds.

The first major reports of ICWs appeared in 1990 and described Ca²⁺ waves propagating through cultures of astrocytes in response to extracellular glutamate (69) or through airway epithelial cells following mechanical stimulation of a single cell (308). Subsequently, ICWs have been found to be initiated by a variety of stimuli in a diversity of cell types, including glial cells (246, 312, 361), neurons (56, 153), various epithelial (171, 249) and endothelial (47, 129, 382) cells, smooth muscle cells (151, 407), cardiomyocytes (352), hepatocytes (122), osteocytes (178), chondrocytes (189), kidney cells (276, 402), mammary gland cells (104), mast cells (255), pancreatic acinar cells (409), and keratinocytes (188). FIGURE 1 illustrates typical examples of ICWs in glial and endothelial cells. While most of these ICWs have been observed in vitro in response to experimental stimuli, it is important to emphasize that ICWs also occur in vivo in organs such as the liver and brain (161, 238, 242, 292, 394, 410). However, the relationship of these in vivo ICWs with respect to normal organ function and physiological stimulation remains to be determined.

ICWs are complex spatiotemporal events that involve active signal transduction within and between cells but are limited by the passive physical characteristics of diffusion and cell architecture. As a result, our understanding of ICWs has arisen from investigations with real-time microscopy, molecular biology, pharmacology, and mathematical modeling. Interestingly, the use of mathematical models to understand intra- and intercellular Ca²⁺ waves has itself driven the further development of the predominantly time-domain models, which addressed mechanisms of Ca²⁺ oscillations, into two- and three-dimensional spatial models. In conjunction with complementary experimental approaches, modeling studies now form an essential tool for exploring the mechanisms and behavior of ICWs. Specifically, models allow us to estimate valid ranges of critical parameters that are difficult to experimentally measure (e.g., the permeability of gap junctions) and, importantly, make predictions that can be experimentally tested. Consequently, we present here an integrated view of ICWs based on results obtained with experimental and mathematical approaches, although most of the major concepts have arisen from original experimental observations.

It is now recognized that there are several key processes that determine the appearance and kinetics of ICWs; these are the mechanisms of 1) Ca²⁺ wave initiation, 2) Ca²⁺ wave propagation within cells (intracellular), 3) Ca²⁺ wave propagation between cells (intercellular), and 4) messenger regeneration for the propagation of the Ca²⁺ waves. We will consider each of these processes in turn.

II. Ca²⁺ WAVE INITIATION

ICWs can be initiated by focal stimulation of a single cell or by the global stimulation of a population of cells with selected chemical ligands or specific extracellular conditions. Focal initiation has been commonly achieved by gentle mechanical stimulation of a cell with a micropipette; this is proposed to result in membrane stress that triggers the production of inositol 1,4,5-trisphosphate (IP₃) in the stimulated cell (33, 57, 306, 307, 387). Most studies utilizing mechanical stimulation have been performed with monolayer cell cultures, but mechanical stimulation has also been applied to more complex tissues; for example, mechanical stimulation of part of the brain isolated from embryonic rats results in large ICWs propagating over almost an entire hemisphere (394).

A disadvantage of mechanical stimulation is that it may lead to plasma membrane disruption; this would allow both Ca²⁺ entry into the cell and the liberation of cell constituents, such as ATP or other messengers, from the cell. The possibility of membrane disruption complicates the mechanistic analysis of any ensuing ICW, especially in studies that manually implement mechanical stimulation where the quantification of both the mechanical stimulus and the associated stimulus intensity-response is difficult. The original studies with airway epithelial cells were performed using a piezo-electric device that allowed the deflection of the mechanical stimulus to be increased (under voltage control) to generate a threshold pulse. Because mechanical stimulation is germane to mechanosensitive cells (e.g., osteocytes that utilize mechanical forces to regulate bone deposition and turnover), a more sophisticated method of mechanical stimulation involving nano-indentation of the plasma membrane with an atomic force microscope (165) has been developed for osteocyte stimulation. On the other hand, strong mechanical stimulation that results in cell injury might be relevant for the investigation of the role of ICWs in wound healing processes (72, 169, 257).

Alternatively, focal stimulation has been achieved by electrical stimulation (mostly in neural tissues) with an extracellular microelectrode (111, 139, 147, 316, 385) or the local release of messengers, such as ATP (7, 143, 316), endothelin-1 (385), glutamate (353, 385), and nitric oxide (NO) (396) by pressure ejection, iontophoresis, or flash photolysis. Focal stimulation can also be achieved at specific stages of the signal transduction cascade of ICWs by a steplike elevation of the intracellular concentration of signaling messengers such as IP₃ or Ca²⁺ (FIGURE 1). This is usually accomplished by an intracellular injection of an inactive caged compound followed by its photolysis to release the active compound or by the focal application of a Ca²⁺ ionophore (33, 173, 210, 221, 261, 409, 415). Microinjection of recombinant Bax, a pro-apoptotic member of the Bcl-2 family of apoptotic proteins, also triggers ICWs, probably via Ca²⁺ release from mitochondria (50). Recent work from Nedergaard's group has demonstrated, in hippocampal brain slices, ICWs triggered by a local decrease in extracellular Ca²⁺ following the photoactivation of the Ca²⁺ buffer diazo-2 (371a). The ICW is initiated in astrocytes by the opening of connexin hemichannels and subsequent release of ATP (see section IVB). These nonmechanical stimuli allow for a better assessment of the stimulus intensity-response (210) and provide the ability to determine if the active messenger concentration is in the physiological, supraphysiological, or pathophysiological range.

Global stimulation can be achieved by applying mechanical stretch to cell monolayers (177), by uniformly exposing cells to a neurotransmitter [e.g., glutamate (57, 69) or serotonin (421)], or by reducing the extracellular Ca²⁺ concentration (11, 413). Interestingly, ICWs can also occur spontaneously without specific triggers. Spontaneous ICWs have been observed in retinal pigment epithelial cells (270), the intact retina in vitro and in vivo (193), brain slices of the developing neocortex (394), Bergmann glia (astrocytes in the molecular layer) of the in vivo cerebellum (161), and hippocampal astrocytes in vivo (192). The frequency of these spontaneous ICWs can be increased by lowering extracellular Ca²⁺ (343, 394, 413), an effect that may result from increased connexin hemichannel opening and ATP release (11) (see sect. IVB).

Unfortunately, the mechanisms initiating all these different ICWs are ill-defined and vary. In general, stimuli that bring about a local increase in [Ca²⁺]_i are often wave-triggering events. However, in early work, it was found that mechanical stimulation in extracellular Ca²⁺-free conditions initiated an ICW even though the stimulated cell did not display an [Ca²⁺]_i elevation (33, 35) (but opposing results have also been reported, Ref. 385). This indicated that an [Ca²⁺]_i increase within the stimulated cell was not essential to trigger an ICW and led to the idea that a messenger other than Ca²⁺ was involved. It is now accepted that this alternative signal is the signaling messenger IP₃, which serves as an agonist of Ca²⁺ release from the endoplasmic reticulum (ER) and underlies dynamic [Ca²⁺]_i changes, including Ca²⁺ oscillations (23). An [Ca²⁺]_i increase by itself can, in some cases, also trigger ICWs. A detailed analysis of the role of IP₃ and Ca²⁺ as wave initiators and messengers of ICW propagation is presented below.

III. INTRACELLULAR Ca²⁺ WAVE PROPAGATION

The mechanism for the propagation of Ca²⁺ waves through cells is strongly dependent on the Ca²⁺ excitability of the cytosol and the diffusion properties of the messenger involved. Ca²⁺ excitability refers to the ability with which a small local transient elevation in [Ca²⁺]_i can be amplified into a large spreading Ca²⁺ pulse or wave. To understand these amplification mechanisms, the characteristics and spatial distribution of the Ca²⁺ release channels involved must first be considered.

A. Ca²⁺ Release Channels

There are two major Ca²⁺-permeable receptor channels that reside within the internal membranes of most cells [usually the ER or sarcoplasmic reticulum (SR) membranes of muscle cells]; these are the IP₃ receptor (IP₃R) and the ryanodine receptor (RyR, named for its affinity for the plant alkaloid ryanodine). In addition, other Ca²⁺-permeable channels including polycystin-2 and two-pore channels can also contribute to [Ca²⁺]_i increases. The properties of these channels are discussed below.

1. IP₃Rs

IP₃Rs are tetrameric assemblies of large subunits (310 kDa) that have six transmembrane domains. Each subunit has a mushroomlike structure; the larger dome (which includes the COOH and NH₂ termini and comprises ∼85% of the protein) projects into the cytoplasm while the stalk corresponds to a transmembrane domain and luminal loop associated with the ER (364). The cytoplasmic domain contains an IP₃ binding site and a regulatory/coupling region (118).

The gating of the IP₃R involves the processes of activation, inhibition, and inactivation. These, in turn, are influenced by the cytoplasmic concentrations of IP₃, Ca²⁺, H⁺, and ATP as well as by the redox and phosphorylation status, protein interactions, and the Ca²⁺ concentration in the ER lumen (118). The binding of IP₃ to the IP₃R increases its open probability, primarily by modifying their sensitivity to inhibition by Ca²⁺ (118). At low IP₃ concentrations, the IP₃R is very sensitive to Ca²⁺ inhibition. At higher IP₃ concentrations, the susceptibility to Ca²⁺ inhibition is reduced and reversed, which leads to increased open probability. However, because there are three different IP₃R isoforms (types 1, 2, and 3), the IP₃ concentrations necessary for channel opening depend on the IP₃ affinity of each isoform, and this can vary from tens to hundreds of nanomolar (118).

The [Ca²⁺]_i dependence of IP₃R open probability has a biphasic, bell-shaped relationship with a central peak at ∼200 nM for all three isoforms (26, 380). At concentrations <200 nM, Ca²⁺ progressively increases the open probability of IP₃R channel that has been sensitized by the binding of IP₃. This increasing sensitivity to Ca²⁺ thereby serves as a mechanism of Ca²⁺-induced Ca²⁺ release (CICR) (293). Ca²⁺ activation of type 1 IP₃Rs is positively cooperative, giving a steep increase in open probability over a narrow [Ca²⁺]_i range and is ideally suited for CICR. Type 3 IP₃R channels have a broader [Ca²⁺]_i activation range and a higher sensitivity to Ca²⁺ activation that is better suited for channel opening in response to small Ca²⁺ changes when the IP₃ concentration is low. A consequence of this sensitivity of individual channels to Ca²⁺ is that it provides a mechanism for the activation and recruitment of nearby IP₃Rs. A Ca²⁺ signal can spatially expand, from a localized Ca²⁺ blip at a single receptor (∼1 μm spread) to a Ca²⁺ puff formed by the cooperative opening of a number of IP₃Rs (∼4 μm spread) to an intracellular Ca²⁺ wave that results in a global [Ca²⁺]_i elevation (22, 36).

In contrast, higher Ca²⁺ concentrations (>200 nM) result in the progressive inhibition of IP₃R opening. This mechanism serves to terminate the Ca²⁺ release process. While this effect appears to be mediated by two independent Ca²⁺-binding sites with different affinities for Ca²⁺ (224), it is not clear if these Ca²⁺ binding sites are part of the IP₃R or an associated accessory protein. Some reports indicate the bell-shaped Ca²⁺ dependency is not universal for all receptor isoforms, with types 2 and 3 channels being insensitive to high [Ca²⁺]_i (140, 283). However, this may result from experimental conditions using purified receptors where accessory proteins are absent.

IP₃Rs are also influenced by the cytoplasmic ATP concentration. Although not required for channel activity, increased ATP concentrations potentiate channel activity by enhancing its sensitivity to Ca²⁺ activation. The effect of ATP is best characterized for types 1 and 3 IP₃Rs. Luminal Ca²⁺ also influences the IP₃R, with high luminal [Ca²⁺] inhibiting channel activity (25, 366).

2. RyRs

RyRs are homotetramers of large (>550 kDa) subunits that form a square structure around a central pore that also mediates and regulates Ca²⁺ release from the SR in muscle cells and ER in nonmuscle cells (201, 412). Like IP₃Rs, they have a large cytoplasmic NH₂-terminal domain that modulates channel gating and acts as a scaffold for interactions with regulatory proteins like calmodulin and FK506-binding protein. The COOH terminus is shorter and is involved in pore delineation within the SR/ER. There are three RyR isoforms: RyR1, RyR2, and RyR3. While the RyR3 is common in the brain and occurs in the spleen, heart, and testis, it is not well characterized. On the other hand, RyR1 and RyR2 are well characterized; RyR1 is predominately expressed in skeletal muscle, but is also found in other tissues, RyR2 is predominately expressed in cardiac muscle and widely expressed in nonmuscle tissues including smooth muscle, adrenal glands, lung, and brain.

Although similar in structure, the activation mechanisms of RyR1 and RyR2 are different. RyR1 is activated by a molecular reconfiguration mediated by the dihydropyridine receptor domain of plasma membrane Cav1.1 channels in response to membrane depolarization. In contrast, membrane depolarization in cardiac muscle activates Cav1.2 channels on t-tubule membranes to stimulate Ca²⁺ entry; this Ca²⁺ directly binds to and opens the RyR2. RyR open probability is also influenced by several other factors including, Mg²⁺, cADP-ribose, ATP, caffeine, ryanodine, FK506-binding protein, and ER/SR luminal Ca²⁺ (reviewed in Refs. 102, 201, 412).

While the initial opening trigger may differ, the open probability of RyRs (like IP₃Rs) is enhanced (namely by CICR) and inhibited by Ca²⁺ in a biphasic, bell-shaped manner (412). For RyR1, enhancement occurs at submicromolar [Ca²⁺]_i by binding to high-affinity Ca²⁺ sites (“A-site”) while inhibition occurs in the 100 μM to 1 mM range by binding to low-affinity sites (“I-site”) (228, 229). RyR2 require micromolar [Ca²⁺]_i for their activation and above 1 mM [Ca²⁺]_i for inhibition (203, 399). For Ca²⁺ wave propagation, this form of Ca²⁺ amplification via CICR through the RyR is important. Ca²⁺ can also influence RyRs indirectly, via Ca²⁺-calmodulin interactions that shift the Ca²⁺ dependence of RyR2 activation to higher [Ca²⁺]_i, thus decreasing the probability of RyR2 opening (14, 400). Mg²⁺ competitively antagonizes Ca²⁺ binding at the A-site and acts in a cooperative manner with Ca²⁺ at the I-site, causing inhibition of CICR (204). Increased [Ca²⁺] in the ER/SR store lumen stimulates Ca²⁺ release via RyRs while a decrease of store [Ca²⁺] inactivates the RyR and contributes to termination of CICR (138, 365). Thus the susceptibility of RyR to CICR is higher upon increased loading of Ca²⁺ stores (38, 339).

As mentioned, a variety of compounds can alter RyR activity. cADP-ribose may act as a RyR activator in nonmuscle cells (230) by sensitizing the RyR to Ca²⁺ (206), but it is unclear if this is a direct or indirect effect. ATP and other adenine nucleotides can potentiate RyR Ca²⁺ release (226, 227). Caffeine also increases RyRs open probability by acting cooperatively with Ca²⁺ and ATP to increase their binding affinity (298). The immunosuppressants FK506 and rapamycin act via the FK506-binding protein (calstabin) to stabilize the closed state of the channel, thereby limiting CICR. The alkaloid ryanodine locks the RyR in an open subconductance state at low (nanomolar) concentrations and inhibits the RyR at high (100 μM) concentrations (412). These opposing effects are mediated by two different ryanodine-binding sites with distinct affinities.

3. CICR

It is clear that both IP₃Rs and RyRs on the ER/SR are equipped with CICR mechanisms. CICR is mediated by a positive-feedback mechanism in which the local increase in Ca²⁺ acts on the receptors to further increase their open probability and enhance Ca²⁺ efflux (FIGURE 2). Although the IP₃R and RyR are distributed throughout the cell (in the ER/SR), their distribution may not be uniform; RyRs can form clusters of at least four to six RyR channels (388) that together generate Ca²⁺ sparks (59), a brief localized increase in Ca²⁺ (∼2 μm spread). Similar clusters of IP₃Rs lead to the generation of Ca²⁺ puffs. Alternatively, RyR and IP₃Rs may form mixed clusters (282) whereby Ca²⁺ increases resulting from IP₃Rs can stimulate adjacent RyRs. In addition, mitochondria may also contribute to CICR (298); increased [Ca²⁺]_mito resulting from the uptake of Ca²⁺ released from the ER can open the mitochondrial permeability transition pore (PTP) which can serve as a mitochondrial Ca²⁺ release pore (291). Additionally, indirect ways of CICR are also possible, for example, via activation of PLC by Ca²⁺ (see sect. VIA).

Figure 2.

Basic mechanisms driving a diffusive Ca²⁺ wave across a cell (either initiating or communicating the wave; see sect. III). Inositol 1,4,5-trisphosphate (IP₃) diffuses across the cell and binds to an IP₃ receptor (IP₃R) to stimulate Ca²⁺ release from the endoplasmic reticulum (ER). The positive feedback (+ve) of Ca²⁺ on the IP₃R initiates Ca²⁺-induced Ca²⁺ release (CICR) through the IP₃R. The elevated [Ca²⁺]_i has a negative feedback (−ve) on the IP₃R to terminate CICR and limit the [Ca²⁺]_i increase. Some Ca²⁺ may diffuse to adjacent ryanodine receptors (RyR) to initiate CICR via these receptors. However, IP₃ diffusion occurs more quickly than Ca²⁺ diffusion (due to cytosol protein buffers) and propagates the Ca²⁺ wave more efficiently. IP₃ may have been produced in response to cell stimulation or may have entered the cell via a gap junction.

4. Other Ca²⁺ release channels

Polycystin-2 channels (PC-2) are members of the TRP channel family (TRPP2) that form Ca²⁺ release channels located on the ER (8, 37). They are activated by Ca²⁺ and may contribute to CICR in concert with IP₃Rs and RyRs (9, 305). Exogenous expression of PC-2 potentiates IP₃-triggered Ca²⁺ elevation (305) and may thus amplify the Ca²⁺ changes associated with ICWs.

Two-pore channels (TPC) are NAADP-gated Ca²⁺ release channels on acidic Ca²⁺ stores (lysosomes and/or endosomes). They do not contribute to CICR, but Ca²⁺ released via these channels may recruit activation of neighboring IP₃Rs and RyRs, thereby generating an intracellular Ca²⁺ wave that transforms a local Ca²⁺ signal into a global Ca²⁺ signal (418). Direct interactions between TPCs and RyRs may further contribute to shaping the Ca²⁺ signal (267). It has been suggested that NAADP might function as a base signal, recruiting IP₃Rs and RyRs (49).

B. Mechanisms of Ca²⁺ Amplification

The mechanism of intracellular Ca²⁺ wave propagation that was first experimentally described was based on the diffusion of IP₃ from a restricted stimulation site. This type of Ca²⁺ wave was hypothesized to be initiated by mechanical stimulation and is reproduced by a transient, intracellular application of IP₃ (by pressure injection, iontophoresis, or photo-activation of inactive caged IP₃ precursors) (33) (FIGURE 1). Following its liberation, IP₃ immediately diffuses in all directions, and in doing so, it encounters, binds to, and activates IP₃Rs to release Ca²⁺ from the ER. As mentioned earlier, the IP₃R is also sensitive to Ca²⁺. Consequently, a self-amplifying increase in [Ca²⁺]_i, mediated by CICR via the IP₃R, follows the diffusive spread or wave of IP₃ (FIGURE 2). Subsequently, the IP₃ signal is terminated by its dephosphorylation to IP₂ by a 5-phosphatase or its phosphorylation to IP₄ by a 3-kinase (323).

Because Ca²⁺ can also activate RyRs via CICR, it is possible that Ca²⁺ wave propagation may be enhanced by local increases and diffusion of Ca²⁺, especially if the IP₃R and RyR are colocalized in receptor clusters. In fact, the gain of CICR determines the Ca²⁺ excitability of the cytoplasm and a larger gain increases the propensity of intracellular Ca²⁺ waves that are based on Ca²⁺ diffusion between Ca²⁺ release sites (38). Thus intracellular Ca²⁺ waves occur more frequently upon increased filling of the Ca²⁺ stores (217), upon exposure to cADP-ribose (401), caffeine (94), or a slight elevation of the intracellular IP₃ concentration (not triggering a [Ca²⁺]_i transient by itself). Thus the rate of intracellular Ca²⁺ wave propagation is a complex function of IP₃ and Ca²⁺ diffusion, sequential CICR via sensitized IP₃Rs and RyRs, and the distribution and composition of the ion channel/receptor clusters (FIGURE 2).

Because CICR through the IP₃R and RyR is a self-limiting process and only continues until the ER/SR is empty or the channels have closed, the [Ca²⁺]_i increase associated with a Ca²⁺ wave often has a similar amplitude at all locations across the cell. In contrast, the concentration of the underlying diffusive IP₃ wave is believed to decrease with distance from the initiation site (proportional to the inverse distance squared for two dimensions) (FIGURE 2). Unfortunately, unlike the changes in [Ca²⁺]_i, changes in [IP₃]_i are less easily visualized by imaging techniques because of the limitations of reporter dyes for IP₃. However, it is important to point out that as long as the [IP₃]_i at the diffusive wave front remains greater than the threshold for activation of the IP₃R (∼10–30 nM), Ca²⁺ release will continue to be initiated and a Ca²⁺ wave will appear to propagate without a degradation. In large cells, such as oocytes (diameter >100 μm), smooth muscle cells, or cardiomyocytes, the Ca²⁺ waves may have a trailing edge of decreasing [Ca²⁺]_i which represents the termination of Ca²⁺ releases and return of most of the Ca²⁺ to the ER by the action of sarcoplasmic/endoplasmic reticulum Ca²⁺-ATPase (SERCA) (FIGURE 2). However, in small cells (diameter 10–20 μm), Ca²⁺ waves often do not have a trailing edge because the period of Ca²⁺ release persists for longer than the time taken for the Ca²⁺ wave to travel across the cell.

IV. INTERCELLULAR Ca²⁺ WAVE PROPAGATION

At the cell border, the cell membrane limits both IP₃ and Ca²⁺ diffusion and hence CICR. Consequently, the transmission of a Ca²⁺ wave to the neighboring cell must occur by a modified or alternative mechanism. There are two major scenarios for this process. The mechanism originally characterized for cell-cell communication occurs via gap junctions. The mechanism subsequently discovered occurs by paracrine signaling with an extracellular messenger (FIGURE 3).

Figure 3.

Overall hypothesis for the communication of ICWs. Local stimulation (see sect. II) can lead to elevated IP₃ in the initiator cell. As IP₃ diffuses across this cell, it generates an intracellular Ca²⁺ wave by the release of Ca²⁺ through IP₃Rs with amplification from RyRs. Diffusion of IP₃ through a gap junction to an adjacent cell initiates a second intracellular Ca²⁺ wave, via IP₃Rs and RyRs. In addition, or alternatively, the stimulated cell releases ATP (in either a Ca²⁺-dependent or -independent manner) via plasma membrane channels (hemichannels, maxi-anion channels) or vesicular release. The extracellular diffusion of ATP (or other messengers; see sect. IVB4) to adjacent cells activates P₂ receptors which, in turn, stimulate IP₃ production to generate a Ca²⁺ signal in the adjacent cell. Propagation may involve the regenerative release of ATP (see sect. VIB). Each mechanism can occur in isolation or synergistically.

A. Gap Junction Communication

Gap junctions are arrays of transmembrane channels that connect the cytoplasm of two adjacent cells and thereby provide a direct diffusion pathway between the cells. Cells sharing a gap junction channel each provide a hemichannel (also known as a connexon) that interacts, head-to-head, to form a complete gap junction channel (302). Each hemichannel is composed of proteins called connexins that exist as various isoforms encoded by 20 different genes in mouse and 21 genes in humans (332). Of these connexin (Cx) isoforms (named Cx23 through Cx62), Cx43 is the most ubiquitous (198). The properties of the connexins forming each gap junction determine the physiological characteristics of the gap junction; these properties include the permeability profile and the gating characteristics. The maximum molecular weight (MW) of compounds that are able to diffuse through gap junctions is in the order of 1,000–1,500. Thus gap junctions are believed to be readily permeable to Ca²⁺ (atomic weight, 40) and IP₃ (MW, 486) (303). However, gap junction permeability for IP₃ differs with connexin type, and this results in the permeability of IP₃ through Cx32 gap junctions being greater than that through Cx43 gap junctions, which in turn is greater than that through Cx26 gap junctions (248).

The idea that ICWs may utilize gap junctions was initially suggested by the kinetics of the propagation of ICWs at cell boundaries in airway epithelial cells. Upon reaching the cell boundaries, these ICWs were delayed for periods of up to 1 second. The ICWs that were subsequently initiated in adjacent cells spread out from localized membrane sites all in contact with the initiating cell. Moreover, these ICWs did not initiate at other cytoplasmic sites. This need for cell contact was tested by examining the propagation of ICWs near discontinuities in the epithelium (usually created by scoring the monolayer). In airway epithelial cells, mechanically stimulated ICWs did not communicate across these spaces (Sanderson, unpublished results). Similarly, these ICWs were not influenced by a fluid flow over the apical surface of cells, whereas ICWs mediated by ATP release (e.g., from a micropipette) were biased in the direction of fluid flow (143).

Because it had been previously found, by freeze-fracture techniques and electrophysiology, that cultured airway epithelial cells possessed gap junctions, the influence of halothane, an inhibitor of gap junction conductivity, on ICW propagation was examined. In the presence of halothane, the initiating Ca²⁺ wave was restricted to the stimulated cell, a result implying a dependence of ICWs on functional gap junctions (308). However, caution is required when using gap junction blockers; a similar inhibition of ICWs has been reported with octanol or heptanol, but it appears that these compounds can also dampen IP₃-mediated Ca²⁺ release or capacitative Ca²⁺ entry (184), effects that would appear similar to blocking the ICW.

Another experimental approach emphasizing the role of gap junctions in ICW propagation consisted of high-resolution imaging of HeLa cells transfected with Cx43 conjugated to green fluorescence protein (Cx43-GFP). These experiments demonstrated that the intracellular Ca²⁺ waves initiated in the receiving cell always started from a site of cell membrane contact with the donor cell that corresponded to the location of the gap junctions (259). In contrast, in ICWs mediated by paracrine ATP signaling (discussed later), the Ca²⁺ signals developing in the receiving cell could occur anywhere within the cell, but often initiated in the region of the perinuclear ER.

Mechanically induced ICWs were also found in cultured glial cells (57), and this led to the study of ICWs in glioma cells that lacked gap junctions (58). In these cells, mechanically stimulated Ca²⁺ waves were restricted to the stimulated cell (this restriction also implies that extracellular ATP is not released by mechanical stimulation mediated with a piezo-electric device). However, when glioma cells were transfected and demonstrated to express Cx43 gap junctions and electrical coupling, ICWs were again observed (FIGURE 1; similar results with Cx26). Furthermore, a correlation between increased connexin expression and increased distance of ICW propagation was observed (58). These data appeared to strongly support the concept of gap junction-mediated ICWs. However, as will be discussed later, the presence of connexins in glioma cells also appears to enhance ATP release. Therefore, with hindsight, these data were not definitive proof that gap junctions were the only route of communication.

The possibility that IP₃ can diffuse between cells to generate ICWs was first indicated by the stimulation of an ICW following the iontophoresis of IP₃ into a single cell of cultured epithelial cells (308) or a group of isolated hepatocytes (303). However, more convincing results were achieved by the microinjection of caged-IP₃ into a single cell of a glial cell culture. As predicted, flash photolysis of the caged-IP₃ within the injected cell initiated an ICW. Moreover, flash photolysis of a distal, noninjected cell, also initiated an ICW (210). The caged-IP₃ used in these studies was not cell permeant. As a result, the most obvious mechanism for caged-IP₃ to reach the distal cells was by diffusion through gap junctions. Caged-IP₃ has a molecular weight slightly higher than that of IP₃; this implies that IP₃ may actually have a higher diffusion rate though the cells.

In keeping with this idea, cells with gap junctions composed of a mutant form of Cx26 (V84L-mutant channels, associated with deafness) that were impermeable to IP₃ but otherwise had normal electrical properties and permeability for Lucifer Yellow (LY, a fluorescent dye with MW 457) failed to produce ICWs in response to IP₃ injection (17). The idea that the propagating ICWs are the result of sequential activation of the IP₃R is also supported by the inhibition of ICWs by the introduction of heparin, an IP₃R antagonist, into contiguous epithelial cells by high-frequency electroporation (33). Collectively, these studies underscore the importance of gap junctions as a conduit for IP₃ between cells.

1. Which messenger: IP₃ or Ca²⁺?

The idea that IP₃ can pass through gap junctions to mediate ICW propagation consistently leads to the proposition that Ca²⁺, because of its smaller size, must also serve as a messenger of ICWs. Indeed, the very appearance of Ca²⁺ waves spreading from one cell to the next suggests a large-scale movement of Ca²⁺. However, the relative importance of each messenger in the propagation of ICWs varies significantly with the characteristics of the cell system and the nature of the stimulation initiating the ICWs. Important cell characteristics affecting propagation include the extent of Ca²⁺ buffering, the distribution and sensitivity of IP₃R and RyRs (contributing to CICR), and the number and type of gap junctions. The mode of stimulation may either be local (single cell) or global (multiple cells), and this dictates if the ICWs are propagating into cells in a quiescent or excited state.

Because Ca²⁺, but not IP₃, is strongly buffered by cytoplasmic proteins, Ca²⁺ movement within a cell is very restricted and therefore considerably slower than that of IP₃ (effective diffusion coefficients of ∼13 and ∼280 μm²/s for Ca²⁺ and IP₃, respectively; Ref. 5). In fact, Ca²⁺ attrition as a result of binding with buffers requires that many more moles of Ca²⁺, compared with IP₃, need to be released from a point source for Ca²⁺ to diffuse over similar distances. Thus IP₃ is more likely to diffuse greater distances at faster speeds, compared with Ca²⁺, from point sources in cells, especially when encountering gap junctions that would greatly reduce the point source concentration of the messenger in the connected cells (64).

The inefficiency of Ca²⁺ to propagate ICWs between cells is also emphasized in cells that are directly connected via tunneling membrane nanotubes; these nanotubes do not contain gap junctions but are long (several tens of micrometers), thin membrane protrusions or tubes that directly connect the cytosol of one cell to a distant neighboring cell (81, 299). Tunneling nanotubes allow for the exchange of organelles and contain ER with IP₃Rs. However, the propagation of a Ca²⁺ signal through these nanotubes cannot be driven by high [Ca²⁺] at one end but can occur in response to increased [IP₃] and CICR via IP₃Rs (327).

Because IP₃ stimulates the release of Ca²⁺ from IP₃Rs, it must be appreciated that a local increase in Ca²⁺ serves as an indicator of IP₃ diffusion and not necessarily a bulk movement of Ca²⁺. As a result, it is more likely that IP₃ diffusion into an adjacent cell will reignite CICR via the IP₃R and/or the RyR to generate the observed Ca²⁺ response associated with the propagation of ICWs from a local stimulation through adjacent but quiescent cells (FIGURE 3).

The mostly likely scenario in which Ca²⁺ can serve as the primary intercellular messenger is in the presence of global agonists that results in a uniform, but low concentration of IP₃, throughout the multiple exposed cells. Under these conditions, a local increase in Ca²⁺ in one cell could initiate an intracellular Ca²⁺ wave by stimulating CICR via IP₃-sensitized IP₃Rs (and RyRs). Ca²⁺ may then diffuse to an adjacent cell to initiate another Ca²⁺ wave. However, Ca²⁺ diffusion will still be restricted by cytosolic Ca²⁺ buffers, and this limitation of Ca²⁺ as an intercellular messenger is highlighted by the common observation that agonist-induced Ca²⁺ oscillations in multicellular systems (cells in culture or tissues that have gap junction communication) generally occur asynchronously. Under these conditions, where IP₃ is elevated throughout the cells, oscillatory increases in [Ca²⁺]_i do not commonly propagate to adjacent cells.

The possibility that Ca²⁺ can serve as a primary messenger is greatly enhanced by increased gap junction cell coupling. This allows greater amounts of Ca²⁺ to reach the adjacent cell and overcome the buffering capacity of the cytoplasm to allow Ca²⁺ to reach an ER receptor. Increased gap junction coupling can be achieved by increasing the permeability or numbers of the specific connexin channels involved. Examples of this type of Ca²⁺-based communication have been reported in pancreatic acinar cells (409) and blowfly salivary gland cells (420). A combination of two diffusive processes may also occur in which IP₃ initially arrives at receptor clusters, but at concentrations too low to trigger significant Ca²⁺ release. However, this IP₃ concentration may be sufficient to sensitize the IP₃Rs to the following Ca²⁺ spike that can initiate CICR (421).

Of note here is the fact that cardiomyocytes which are highly coupled by gap junctions and have a high density of RyRs (∼100-fold greater than the density of IP₃Rs) (185) only rarely display “slow” ICWs (199) (propagating at a velocity of 10–20 μm/s). Normally, these cells display much faster propagating increases in Ca²⁺ as a result of propagating membrane action potentials that activate membrane voltage-dependent Ca²⁺ channels and RyRs (via CICR). The result of such electrical pacing is that the Ca²⁺ content of the SR of cardiomyocyte is repetitively lowered and the RyR sensitivity to CICR is dampened. This, together with facts that the RyRs are located at some distance from the gap junctions (183) and that mitochondria, which buffer Ca²⁺ (216), are in close apposition to the intercalated disks (91), guards against the occurrence of spurious ICWs mediated by Ca²⁺ diffusion. However, when RyR sensitivity to Ca²⁺ is increased by Ca²⁺ store over-loading (either by β-adrenergic stimulation or increased Ca²⁺ influx), the propagation of spontaneous ICWs increases (199, 234) because of the greater chance that Ca²⁺ diffusion can initiate CICR. Under certain experimental conditions (i.e., cell culture), cardiomyocytes may propagate ICWs mediated by IP₃ diffusion (352); this may be facilitated by the fact that IP₃Rs are located more closely to the gap junctions (intercalated disks) than RyRs.

Mathematical modeling of ICWs (in 1 and 2 dimensions) in epithelial cells supported the idea that IP₃ diffusion through gap junctions explains the kinetics of ICW propagation (328, 330). The key point addressed by these models was whether ICWs could be driven by IP₃ diffusion if IP₃ was generated only in the stimulated cell; IP₃ regeneration in adjacent cells was excluded. This hypothesis accounts for why ICWs in epithelial cells have a limited propagation distance and appear to terminate at borders of specific cells. The modeling indicated that for ICW propagation to occur over about three to five cells, it was necessary for ∼3 μM IP₃ to be generated in the stimulated cell if the IP₃R sensitivity to IP₃ was on the order of 30 nM. A feature of the initial models was that Ca²⁺ diffusion between cells was not required. However, diffusion of both IP₃ and Ca²⁺ from one cell to the next has been included in later models.

Alterations in gap junction permeability, modeled as reduced diffusion rates of IP₃/Ca²⁺ between cells, modified the propagation distance of the ICWs. There appears to be a minimal gap junction permeability required to allow ICW transmission between cells. Additionally, the gap junction permeability required for the propagation of ICWs is inversely related to the effective diffusion constant of the propagating messenger and the Ca²⁺ excitability of the cytosol (329, 330). Surprisingly, when Ca²⁺, instead of IP₃, is assumed to be the primary messenger, the models predict that the critical gap junction permeability to obtain ICWs decreases with a decreasing effective Ca²⁺ diffusion coefficient. In other words, limited Ca²⁺ diffusion may be beneficial for ICW propagation (158). The explanation for this counterintuitive idea is that limited Ca²⁺ diffusion results in a higher localized Ca²⁺ concentration at the mouth of the gap junction in the adjacent cell and that this increases the chance of reigniting CICR. This scenario requires that the CICR sites (RyRs or IP₃Rs) are close to the gap junctions. In addition, several other factors influence the critical gap junction permeability necessary to allow ICWs, including the CICR gain and the rate of Ca²⁺ sequestration by SERCA.

Further support for the idea that a diffusive wave of IP₃ mediates ICWs was provided by the finding that glial cells, some distance from a stimulated cell, displayed Ca²⁺ oscillations (57). Modeling predicts that the IP₃ concentration in these cells is in the correct range to support Ca²⁺ oscillations. This concept was subsequently confirmed by examining the responses of an individual glial cell to ICWs initiated in other cells at different distances from the observed cell (331, 344). When the initiating cell was close or far away, the Ca²⁺ increase was transient in keeping with the expected high or low diffusion concentration of IP₃. However, at intermediate distances, the [Ca²⁺]_i oscillated, reflecting an intermediate IP₃ concentration compatible with the prediction that Ca²⁺ oscillations can only occur between bifurcated curves of IP₃ concentration dependency (98, 393).

In addition to the diffusion of IP₃ or Ca²⁺ between cells, cADP-ribose (MW, 541), an intracellular messenger that can sensitize RyRs to Ca²⁺, may mediate gap junction-propagated ICWs in lens cells (62) and astrocytes (211). However, these ICWs have slower kinetics and propagate shorter distances compared with ICWs based on IP₃ diffusion.

At the other end of the spectrum, with respect to propagation speed, are fast ICWs that propagate between coupled cells in the order of 100 μm/s. While cardiomyocytes may be the best example of this, similar fast ICWs occur in vascular smooth muscle cells in response to the local application of phenylephrine; ICW propagation results from the passage of electrical current through gap junctions to depolarize cell membranes with the subsequent opening of voltage-operated Ca²⁺ channels coupled with CICR. The latter process is the rate-limiting step in the propagation of this specific kind of ICWs (186). Rapidly spreading ICWs have also been reported in arteriolar endothelial cells in vivo in transgenic mice expressing the Ca²⁺-sensitive probe GCaMP2 in endothelial cells under control of endothelial Cx40 (362). Micro-iontophoresis of acetylcholine was used to trigger the ICWs in endothelial cells, and these propagated through the endothelial cells along the vessel with a velocity of ∼116 μm/s, reaching distances up to ∼1 mm. These endothelial ICWs are associated with vasodilation and are hypothesized to regulate blood flow to the parenchyma by inducing upstream dilation of arterioles (87).

2. Does elevated [Ca²⁺]_i inhibit intercellular Ca²⁺ wave propagation via gap junctions?

A paradox that frequently raises concern is how ICWs continue to propagate when an elevation in [Ca²⁺]_i is reported to block gap junctions. Additionally, phosphatidylinositol 4,5-bisphosphate (PIP₂) depletion (381) [resulting from phospholipase C (PLC) activation and IP₃ production] and protein kinase (PKC) activation (103) also block gap junction communication. The idea that increases in [Ca²⁺]_i may limit ICWs was tested in airway epithelial cells by mechanical stimulation in the presence of a global agonist. Upon exposure to extracellular ATP, airway epithelial cells displayed asynchronous Ca²⁺ oscillations, events that increase Ca²⁺, diacylglycerol (DAG) (to activate PKC) and reduce PIP₂. However, mechanical stimulation of a single cell still initiated an ICW (105), indicating that gap junction communication remained sufficient to allow ICWs. One explanation is that the [Ca²⁺]_i attained during the propagation of an ICW is significantly less than that necessary to block gap junctions, although it could be argued that [Ca²⁺]_i may attain higher values within Ca²⁺ microdomains near the gap junction channel (372). The [Ca²⁺]_i reported to result in inhibition of gap junctions varies widely, from ∼300 nM (73, 205, 218) to micromolar concentrations (297, 335), indicating there is not a single fixed threshold concentration for inhibition. In fact, the sensitivity to [Ca²⁺]_i depends on the animal species (73), the cell type, and the connexins involved. Dakin et al. (78) reported that the source of [Ca²⁺]_i elevation was also important, with capacitative Ca²⁺ entry causing uncoupling while ionomycin-based [Ca²⁺]_i elevation was ineffective.

A second explanation for sustained ICW propagation lies in the temporal domain. If ICWs are mediated by IP₃ diffusion, the diffusion wavefront of IP₃ must run in advance of the increase in [Ca²⁺]_i; hence, any Ca²⁺ effects on gap junctions would occur after the passage of IP₃. A similar scenario is applicable to ICWs mediated by electrical changes where the propagating voltage moves rapidly compared with the Ca²⁺. Furthermore, an [Ca²⁺]_i elevation of several minutes is often required to close gap junctions (194, 205), which is much longer than the time required for a typical ICW to travel over 100 μm (5–10 s at 10–20 μm/s). In line with this, Churchill et al. (63) reported that gap junction coupling between lens epithelial cells was not inhibited by the passage of an ICW but was inhibited by a sustained (several minutes) elevation of [Ca²⁺]_i brought about by the Ca²⁺ ionophore ionomycin. Gap junction closure by [Ca²⁺]_i elevation is thought to be mediated by Ca²⁺-calmodulin signaling (218, 272) and alteration of chemical gating of gap junction channels (their closure by high [Ca²⁺]_i or low pH_i) is, for unknown reasons, a process that takes several minutes (272). The closure of gap junctions by alterations in DAG or PIP₂ levels or PKC activity is unlikely to occur in the nonstimulated adjacent cells propagating an ICW because these cells only respond to IP₃ diffusion from the stimulated cell. However, gap junction closure in the stimulated cell, where IP₃ and DAG are formed from PIP₂, could shut-off the diffusion gradient derived from the stimulated cell and thereby reduce wave propagation distance; the extent of this effect would depend on the time delay until gap junction closure.

On the other hand, the inhibition of gap junctions does not necessarily mean that ICW propagation must be limited. Modeling studies have suggested that the partial closure of gap junctions may in fact result in larger waves (125). This contrarian response is predicted to be the result of a greater regeneration or build-up of IP₃ in each cell (see sect. VIA) due to lower losses of IP₃ or Ca²⁺ through partially closed gap junctions. As a result, high [IP₃] within cells will ensure IP₃ diffusion to neighboring cells and produce a [Ca²⁺]_i elevation despite the lower degree of gap junction coupling.

Lastly, although [Ca²⁺]_i elevation may under certain circumstances inhibit gap junctions and wave propagation, it is now clear that other mechanisms based on paracrine communication via extracellular messengers contribute to ICW propagation and that these can circumvent the gap junction pathway (discussed below).

B. Paracrine Communication

ICWs may also be propagated by paracrine signaling between adjacent cells. In this process, a signaling molecule, released by one cell, diffuses across the extracellular space between cells and, after binding to membrane receptors on neighboring cells, triggers a further elevation in [Ca²⁺]_i (FIGURE 3). Early evidence for this mechanism came from the ability of ICWs to propagate across a cell-free zone and from the observation that an extracellular fluid flow could bias the direction of ICW propagation in the direction of flow (147). In contrast, ICWs relying on gap junctions are not influenced by an extracellular fluid flow (143). However, some caution is required when considering cells that have apical tight junctions; the release of an extracellular messenger into the interstitial space may, in this location, not be influenced by fluid flow across the apical surface of the cells. Sampling of the culture medium from cells after the initiation of an ICW revealed the presence of ATP and the local application of this sampled medium to cells using a micropipette triggered an ICW (137). It has now been demonstrated that ATP is a major extracellular messenger utilized by many cell types, including osteocytes (178), various epithelial (160) and endothelial (128, 370) cells, various cells in the organ of Corti in the inner ear (221, 277), glial cells (137), hepatocytes (317), keratinocytes (379), mast cells (255), and prostate cancer cells (309, 320). Many modeling approaches have also been developed to better understand paracrine ATP-based ICW propagation (18, 19, 166, 220, 337, 379, 392).

1. Mechanisms of ATP release

Advanced imaging techniques in combination with bioluminescent probes that detect ATP have revealed that ICWs are associated with a “cloud” of extracellular ATP (11, 245, 371a). After mechanical stimulation, ATP concentrations were ∼78 μM at the stimulation point but declined to ∼7 μM at a distance of 100 μm away (245). The propagation of the increase in extracellular ATP away from the stimulation point occurred at a speed faster (∼41 μm/s) than the propagation of the associated ICW (∼28 μm/s) (245). This indicates that the changes in [Ca²⁺]_i lag behind the changes in extracellular [ATP], a finding consistent with modeling studies that predict delays inherent to a multistep diffusion-reaction scenario (19). The signaling steps include ATP release and diffusion; the sequential activation of purinergic receptors, G proteins, and PLC; the production of IP₃; and the opening of IP₃Rs to release Ca²⁺ from the ER.

While it is clear that ATP is released during ICWs, the mechanisms mediating ATP release remain controversial, mainly because there are numerous pathways available for ATP release; these include various ATP transporter proteins, vesicular discharge (39; reviewed in Ref. 280) and diffusive flow via ion channels or hemichannels (7). In addition, both Ca²⁺-dependent and Ca²⁺-independent ATP release has been reported (390). Several ATP release mechanisms are sensitive to mechanical stimulation or cell swelling, suggesting they may be directly activated by membrane stress in a Ca²⁺-independent manner. The diversity of ATP release pathways furthermore suggests there will be differences in ICW kinetics. In non-neuronal cells, vesicular ATP release typically occurs with a time scale of seconds (39) while nonvesicular release may be slower, taking minutes to occur (114). Additionally, several ATP release mechanisms may act in combination (280, 286) as well as a regenerative mechanism called ATP-induced ATP release (6) (see sect. VIB). Transporter proteins like the mdr-1 gene product (2) and the CFTR protein (287) have been demonstrated to behave as channels that pass ATP. Both proteins are present in astrocytes (13), while CFTR is found in a wide range of cell types including respiratory tract epithelial cells (253).

In neuronal cells, ATP release is typically vesicular in nature, often occurring in conjunction with the release of other neurotransmitters, but in nonneuronal cells, ATP release may occur by both vesicular and nonvesicular pathways (31, 237). Vesicular ATP release occurs in astrocytes (262, 416), endothelial cells (30), hepatocytes (112), osteoblasts (294), and retinal pigment epithelial cells (286). The identity of the vesicular pool that is released varies but includes quinacrine-labeled vesicles (30, 262, 294), bafilomycin A1-sensitive vesicles (112), or lysosomic vesicles (416).

Diffusive ATP release has been suggested to occur via a maxi-anion channel of unknown identity (301), by hemichannels composed of various connexin isoforms (including Cx26, Cx30, Cx32, Cx36, and Cx43) (7, 109, 179, 318, 326, 343, 374, 417) or pannexins (15, 162, 284, 336, 397), or by the combined action of these channels (121, 212). ATP release via connexin and pannexin hemichannels during ICW propagation has been reported to be Ca²⁺ dependent (83, 86, 215). The connexin hemichannels involved represent the fraction of hemichannels that have not paired-up to form complete gap junction channels and therefore reside in the cell membrane unopposed as large-conductance channels that are maintained in the closed state by a high (millimolar) extracellular Ca²⁺ concentration (107). These hemichannels can be opened by depolarization, alterations in their phosphorylation or redox status, and an [Ca²⁺]_i elevation (67, 83, 86, 289). The pannexins involved have a topology that is similar to connexins but have no primary sequence homology. Importantly, their extracellular loops are glycosylated, and it is believed that this hinders the hemichannel interaction and the formation of pannexin gap junctions (74). There is some evidence for gap juction formation in cells overexpressing Panx1 (15b, 45a, 197a) or Panx3 (171a). The latter study also demonstrated ICWs mediated by Panx3 gap junctions. However, the pannexin family of channels has best been characterized as an ATP release pathway (171a, 15a, 215). Pannexin hemichannels composed of Panx1 have been reported to be the large-conductance pore recruited upon P₂X₇ receptor activation (271). Thus P₂X₇ receptors are equipped with both an ATP receptor as well as an associated ATP-release mechanism that together could contribute to ICW propagation (351). Finally, hemichannels are also equipped with inactivation mechanisms; for connexin hemichannels, channel closure occurs when [Ca²⁺]_i rises above 500 nM (83, 86), and for pannexin hemichannels, channel closure occurs as a consequence of ATP-induced inhibitory feedback (281).

2. Mechanisms of ATP action

While the released ATP can activate both ionotropic (ligand-gated ion channels, i.e., P₂X family) and metabotropic receptors (i.e., P₂Y family), ICWs commonly appear to rely on the stimulation of metabotropic receptors. This involves the activation of G protein-coupled receptors, stimulation of PLC-β, and the generation of IP₃ followed by Ca²⁺ release from the ER via the IP₃R (FIGURE 3). Of the eight identified human P₂Y receptors, five utilize the G protein G_q/11 to activate PLC-β (1). These receptors are the P₂Y₁, P₂Y₂, P₂Y₄, P₂Y₆, and P₂Y₁₁, and all have distinct affinities for nucleotide binding. The activation order for the P₂Y₁ receptor is ADP ≥ ATP; for the P₂Y₂ receptor, UTP = ATP; for the P₂Y₄ receptor, UTP > ATP; for the P₂Y₆ receptor, UDP > UTP > ADP; and for the P₂Y₁₁ receptor, ATP > ADP. Current data implicate the involvement of P₂Y₁, P₂Y₂, and P₂Y₄ receptors in ICWs and, as might be expected, the properties of the purinergic receptors involved are likely to influence the propagation properties of ICWs. For example, ICWs propagating via P₂Y₂ receptors appear to travel faster and further than ICWs propagating via P₂Y₁ receptors (119); this concept is supported by modeling studies (19).

Extracellular ATP is prone to rapid degradation to ADP and AMP by several classes of ecto-nucleotidases. Ecto-nucleoside triphosphate diphosphohydrolases (E-NTPDase 1–8) degrade tri- and dinucleotides while ecto-nucleotide pyrophosphatase/phosphodiesterases (E-NPP 1–3) degrade tri-, di-, and mono-nucleotides yielding adenosine. Degradation of mono-nucleotides to adenosine is also catalyzed by ecto-5'-nucleotidase. The degradation kinetics to adenosine are characterized by a half-life (t_½) on the order of 200 ms (for the hippocampus) (97). A consequence of this rapid degradation of extracellular ATP is that while ICWs may be initially mediated by ATP acting on P₂Y₂ receptors, their propagation may be sustained by ADP activating P₂Y₁ receptors. This multiagonist/receptor combination appears to propagate ICWs further than waves relying only on ATP and P₂Y₂ receptors (119). The experimental presence of the nucleotidase apyrase (to induce rapid breakdown of ATP) blocked ICWs mediated by P₂Y₂ receptors but accelerated ICWs based on P₂Y₁ receptors (ADP generated from ATP) (119). Thus the paracrine contribution to ICW propagation is a complex function of the rates of ATP release and metabolism, the spectrum of purinergic receptors present, and their varying affinities for ATP and its metabolites.

3. Cross talk between connexins and purinergic signaling

After the initial reports that ICWs involve both gap junctions and purinergic signaling, mediated in many cases by ATP release via hemichannels, it was discovered that components of the purinergic signaling cascade were influenced by the expression of connexins. For example, suppression of Cx43 in spinal cord astrocytes induced a change from an ADP-sensitive P₂Y₁ to an UTP-sensitive P₂Y₄ receptor subtype (314, 349). Conversely, overexpression of Cx43 increased the size of ICWs (71) by enhancing ATP release via Cx43 hemichannels rather than by increasing gap junction coupling (70). The influence of Cx43 on P₂Y₁ receptors may be mediated via signaling through the COOH-terminal domain of Cx43 (311).

4. Alternative extracellular messengers

Although ATP appears to be the most common paracrine messenger, ICWs may also be communicated between glial cells by diffusion of extracellular glutamate (93, 168, 263). NO release and downstream PKG signaling are also proposed for paracrine ICW propagation (396). Extracellular Ca²⁺ may function as a paracrine messenger by acting on membrane-bound Ca²⁺-sensing receptors (156). When cells recover from an [Ca²⁺]_i elevation, most cytosolic Ca²⁺ is resequestered into the ER by SERCA, but some Ca²⁺ is pumped out of the cell by plasma membrane Ca²⁺-ATPase (PMCA). Because of the small volume of the interstitial space between cells, this can result in a localized elevation of extracellular [Ca²⁺] that can stimulate Ca²⁺-sensing receptors of neighbor cells, and thereby trigger PLC activation, IP₃ formation and an ICW (155). Modeling studies have further analyzed this specific kind of paracrine ICW propagation (182).

C. Spatial Characteristics of Gap Junction and Paracrine Intercellular Ca²⁺ Waves

In addition to the observation that gap junction-based ICWs typically experience a delay in propagating from one cell to the next (419), these ICWs frequently follow a convoluted pathway through multiple cells because of heterogeneities in the spatial distribution of the gap junctions and the degree of cell-cell coupling. In contrast, paracrine-based ICWs often propagate in a more homogeneous manner characterized by a circular wavefront, but may also be influenced by heterogeneities in the densities of plasma membrane receptors. It is not uncommon for individual cells, probably because of a lack of appropriate receptors, to be bypassed and not exhibit a Ca²⁺ wave, a behavior reiterating the ability of paracrine-based ICWs to propagate across cell-free zones.

Interestingly, despite these differences in propagation pathways, the propagation velocity of gap junction- or paracrine-based ICWs is similar (10–20 μm/s). This is presumably because the molecular weights and therefore the nonbuffered diffusion constants of IP₃, ATP, and glutamate (486, 507, and 292, respectively) are also similar.

D. Combined Gap Junction and Paracrine Intercellular Ca²⁺ Waves

Although we have addressed the propagation mechanisms of gap junction- and paracrine-based ICWs separately, it is important to emphasize that, in many cells, these two mechanisms are simultaneously active and may act synergistically (7, 42, 270, 317, 370). Suadicani et al. (350) demonstrated, by individual or combined overexpression of P₂Y receptors and Cx43, that ICWs utilizing both gap junctions and purinergic signaling have a more coordinated appearance and are less “saltatory” than waves utilizing paracrine signaling only (350). Thus a propagating ICW can be driven by IP₃ diffusing via gap junctions, while ATP is released via Ca²⁺-dependent or -independent mechanisms. ICWs that rely exclusively on IP₃ diffusion via gap junctions propagate only limited distances (∼up to 5 cells), although this depends on the amount of IP₃ generated and the degree of gap junction coupling. In contrast, ICWs relying on paracrine mechanisms propagate greater distances; this is explained by the hypothesis that paracrine signaling often involves a regenerative process to amplify messenger (ATP) release (discussed later) (FIGURE 3). However, stochastic two-dimensional simulations of an ICW have suggested that the combination of gap junction and purinergic signaling is sufficient to generate ICWs with large propagation distances without the need for messenger regeneration (166). The exact contribution of gap junctions or paracrine mechanisms to ICW propagation also appears to correlate with cell differentiation since ICWs in osteoblast-like cells shifted from being predominantly mediated by P₂Y receptors to being mediated by gap junctions as cells differentiated (152). The contribution of gap junctions and paracrine signaling to ICW propagation between different cell types obtained from in vitro cell culture studies is summarized in TABLE 1.

Table 1.

Intercellular Ca²⁺ wave propagation between different cell types

Cell Types Involved	Propagation Mechanism	Reference Nos.
Astrocytes to neurons	Gap junction-based	244
Astrocytes to neurons	Glutamate-based paracrine signaling	263
Astrocytes to neurons	ATP-based signaling	371a
Astrocytes and endothelial cells	Gap junction- and ATP-based signaling	41, 210, 257
Astrocytes and menigeal cells	Gap junction- and ATP-based signaling	135
Astrocytes and microglial cells	ATP-based signaling	316, 386
Astrocytes to Müller cells	ATP-based signaling	245
Astrocytes to oligodendrocytes		266
Satellite glial cells (SGC) and neurons (N) in trigeminal ganglia	SGC to N: gap junction- and ATP-based; N to SGC: ATP-based signaling	348
Hepatocytes to bile duct epithelial cells	ATP-based signaling	317
Vascular smooth muscle cells and endothelial cells	Gap junction-based	170
Supporting cells in the cochlea trigger Ca2+ spikes in inner hair cells in explant cultures	ATP-based signaling	376, 377

All examples listed refer to cell culture studies.

E. Intercellular Ca²⁺ Waves and Ca²⁺ Oscillations

As previously mentioned, a consequence of IP₃ diffusion through multiple cells during ICW propagation is the initiation of Ca²⁺ oscillations. In glial cells, these Ca²⁺ oscillations occurred at spatial locations that reflect optimum IP₃ concentrations within cells to sustain Ca²⁺ oscillations (57, 344) and because the [IP₃]_i determines the Ca²⁺ oscillation frequency, the Ca²⁺ oscillations have unique formats in different cellular zones (331). This behavior may provide cells with specific spatial cues to stimulate an appropriate response to the passage of an ICW. The Ca²⁺ oscillations themselves may also generate coordinating pulses of ATP leading to ICWs (344) or become phase-locked between cells via gap junction coupling (134, 146, 157, 310, 421). The passage of messengers via gap junctions is known to influence the frequency of the Ca²⁺ oscillations and the number of cells displaying Ca²⁺ oscillations (110, 310, 338). On the other hand, strong gap junction coupling appears to curtail glucose-induced Ca²⁺ oscillations in the pancreas to reduce insulin secretion (378). This effect probably results from the tendency for [Ca²⁺]_i to equalize in communicating cells which disturbs the Ca²⁺ oscillation dynamics of the individual cell. Additionally, purinergic paracrine communication is known to influence Ca²⁺ oscillations by exerting a synchronizing influence (150, 187).

V. PASSIVE AND ACTIVE WAVES

ICWs that exclusively rely on the production of IP₃ and its diffusion via gap junctions or on the release of ATP and its diffusion in the extracellular space are termed diffusive Ca²⁺ waves. They are based on a messenger concentration gradient that decreases away from the stimulation point, in much the same way as an electrotonic spread of electrical charge over a length of membrane or cells. As long as the messenger concentration is above the threshold for activation of the target receptor, an [Ca²⁺]_i increase is triggered. Thus single cell stimulation establishes a cytoplasmic or extracellular messenger concentration gradient over various rows of cells, creating the appearance of a cell-to-cell propagating Ca²⁺ signal. Diffusive Ca²⁺ waves are also called passive Ca²⁺ waves; these waves have a wavefront velocity that decreases with distance from the stimulation point because the difference in messenger concentration between two adjacent cells also decreases away from the stimulation point. When the wave propagation front reaches a point where the messenger concentration is approaching the threshold concentration for receptor stimulation, the [Ca²⁺]_i transients will disappear and the wave will stop. Diffusive ICWs provoked by a steplike increase in IP₃ in a single cell often have a limited degree of travel.

When this passive mechanism of propagation is combined with a mechanism of messenger regeneration, the wave propagation becomes active. Active ICWs propagate further than ICWs based on a purely passive/diffusive mechanism. Moreover, the wavefront displays a more constant propagation velocity because the underlying messenger is regenerated in (or outside) each cell and thereby maintains a more constant concentration difference between adjacent cells. As a consequence, the amplitude of the [Ca²⁺]_i transients is expected to be rather constant in cells at progressively further distances from the stimulation point. Theoretically, if messenger regeneration is complete within each cell contributing to the wave, the wave would propagate until it encounters a physical boundary. This propagation mechanism reflects that of action potentials which propagate in an all-or-nothing manner (no loss of amplitude) along meter-long axons. However, ICWs are often based on a combination of passive and active propagation, and the waves arrest at discrete points. In mathematical terms, purely active waves are called traveling waves. Because the degree of involvement of messenger regeneration in ICW propagation is difficult to measure in biological experiments, mathematical modeling is frequently used to estimate regeneration parameters to provide a “best fit” with the experimental data. This approach has led to important insights such as the concept of “limited regenerative signaling” that demonstrates that a limited degree of messenger regeneration is sufficient to produce ICWs that are comparable in size to those experimentally observed and that regenerative mechanisms exert an important regulation of wave propagation (159).

VI. MESSENGER REGENERATION IN ACTIVE INTERCELLULAR Ca²⁺ WAVES

A. Internal Messengers

The generation of IP₃ by PLC-β, PLC-γ, and PLC-δ is a Ca²⁺-dependent process, but the requirement for Ca²⁺ (K_d) differs between PLC subtypes; PLC-β and PLC-γ have a low Ca²⁺ requirement (285), and these enzymes can be active at resting [Ca²⁺]_i. In contrast, the K_d of PLC-δ1 for Ca²⁺ is on the order of 1 μM and is thus activated by micromolar [Ca²⁺]_i. PLC-δ activity is also activated by PIP₂ but inhibited by IP₃ (285). As a result, PLC-δ is proposed to contribute to IP₃ regeneration during ICW propagation; IP₃ entering a cell via gap junctions triggers an [Ca²⁺]_i increase that, in turn, would activate PLC-δ and IP₃ synthesis to enhance ICW velocity and propagation distance by maintaining an elevated IP₃ concentration difference between adjacent cells. Presumably, the elevated IP₃ concentration can also serve as a negative-feedback mechanism to limit PLC-δ activity. The extent to which IP₃ regeneration by PLC-δ contributes to ICW propagation is not well understood, but the inclusion of a limited regenerative process in mathematical models of ICWs often improves the fit to experimental data (159). ICWs based on IP₃ regeneration have been described in clusters of cells in nonconfluent cultures of vascular endothelial cell (95).

B. External Messengers

An early observation made with paracrine-based ICWs that crossed cell-free zones was that the width of the cell-free zone had little effect on ICW propagation in the distal cells (147). This implied that the concentration of the extracellular messenger on the far side of the cell-free zone was rejuvenated regardless of where the ICW was initiated. Most evidence for the regeneration of an extracellular messenger pertains to ATP. The observation that suramin (a nonspecific blocker of P₂ receptors) decreased the size of the extracellular ATP wave (245) led to the proposal that ATP regeneration was involved (193). In addition, the relatively small amount of intracellular ATP within a single cell and the relatively small fraction of this ATP that is released by the stimulated cell both indicate a need for ATP regeneration to drive extensive ICWs (392).

A Ca²⁺-mediated process is an obvious way to regulate ATP regeneration, and this may take the form of Ca²⁺-dependent ATP release via connexin or pannexin hemichannels (83, 86, 215). At the present time, it is not clear if vesicular ATP release is a Ca²⁺-dependent process (see sect. IVB). In contrast, a Ca²⁺-independent mechanism of ATP-induced ATP release mediates ATP regeneration in astrocytes (6). This mechanism, detected with radioactive-labeled ATP in combination with P₂ receptor inhibitors, involved undefined P₂ receptors (EC₅₀ for activation ∼144 μM) and is consistent with a suramin-induced decrease in ATP wave size (245). Although multiple P₂ receptors may be involved, P₂X₇ receptors that are associated with large ATP-permeable pores or release channels, perhaps consisting of Panx1 hemichannels (167, 271), are likely key components. Other ionotropic ATP receptors (e.g., P₂X₁ and P₂X₄) have also been linked to the opening of Panx1 hemichannels (397).

On the other hand, ATP regeneration may not occur in all systems. Real-time bioluminescence imaging of ATP (detected with luciferin/luciferase activity) in various cells (rat and human astrocytes, human bronchial epithelial cells, and human umbilical vein endothelial cells) indicated that, in the absence of extracellular Ca²⁺, ICWs were associated with ATP release from a single cell (11). A burst of ATP from a single cell was implicated by the finding that only one cell accumulated an extracellular fluorescent dye (propidium iodide) that can permeate connexin hemichannels; these data infer hemichannel opening provides a bidirectional permeable pathway allowing the movement of ATP out of the cell and dye uptake into the cell. Similar observations have been made for spontaneous ICWs in retinal pigment epithelial cell cultures (270); only the initiating cell of the ICW displayed propidium iodide dye uptake. A caveat of this hypothesis of a single-cell source for ATP is that a regenerative ATP release occurring via pathways other than hemichannels would not be detected by propidium dye entry.

Lastly, because ATP can be released from cells by distinct pathways, it is possible that ICW propagation may result from a combination of ATP regenerative mechanisms. The initiating stimulus of the ICW may also determine if ATP regeneration is involved. Again, mathematical models of point source ATP release with partial regeneration have led to the conclusion that a combined signaling mechanism that includes partial ATP regeneration provides a better fit with the experimental results (220).

VII. PHASE WAVES

A unique form of Ca²⁺ signaling, characterized in hepatocytes, that appears as an ICW is a phase wave, which is an ordered phase-shifted temporal sequence of independent Ca²⁺ oscillations. If a row of cells is exposed to conditions that generate an increase in IP₃ concentration that progressively decreases in each cell, the ensuing oscillatory Ca²⁺ signals create the impression of a cell-to-cell propagating wave because each cell shows a Ca²⁺ spike with a slight delay with respect to its upstream neighbor (100). In the liver, radial rows of hepatocytes have a gradient of plasma membrane G protein-coupled receptors responsive to the hormones norepinephrine or vasopressin. Uniform exposure to these hormones establishes an IP₃ concentration gradient within the cells. However, this gradient can be modulated by gap junction communication. ICWs based on a phase wave mechanism occur both in vitro and in vivo (within intact perfused liver; see sect. IXE).

VIII. PHARMACOLOGICAL MANIPULATION OF INTERCELLULAR Ca²⁺ WAVES

Almost all of the aforementioned mechanisms of ICWs in isolated cell systems have been elucidated with the extensive use of a variety of pharmacological agents. With this knowledge, we now have a viable “tool kit” to better explore the mechanisms and roles of ICWs in vivo. With this in mind, we have reviewed these pharmacological tools in terms of their action, but, as with most experimental compounds, their nonspecific effects must be considered and data obtained with these agents interpreted with care.

These agents are classified as 1) inhibitors of gap junction coupling, 2) inhibitors of vesicular and nonvesicular ATP release, 3) ATP degrading enzymes, 4) purinergic receptor antagonists, and 5) inhibitors of intracellular Ca²⁺ signaling. TABLE 2 gives an overview of the most frequently used compounds and their targets; gap junction inhibitors are reviewed in detail in References 29 and 334, while References 174 and 251 provide detailed overviews on purinergic receptor antagonists. In addition to the inhibition of ICWs with substances acting at defined targets, a variety of agents are known to either inhibit or promote ICWs as a result of combined effects at several targets. These compounds with modulatory effects are summarized in TABLE 3.

Table 2.

Pharmacological compounds used to inhibit intercellular Ca²⁺ waves

Molecule or Condition	Target Mechanism	Reference Nos.
Halothane	Gap junction channel/hemichannel inhibition	117, 153, 236, 308
Long-chain alcohols (octanol, heptanol, and others)	Gap junction channel/hemichannel inhibition+	56, 117, 219, 236, 240, 245, 352, 379, 402, 407, 415, 421
Glycyrrhetinic acid/carbenoxolone	Gap junction channel/hemichannel inhibition*	40, 42, 43, 136, 153, 164, 192, 214, 243, 266, 269, 275, 295, 296, 355, 358, 370, 371, 379, 384, 385
Fenamates	Gap junction channel/hemichannel inhibition+	127, 153
Connexin mimetic peptides like Gap26 or Gap27	Gap junction channel/hemichannel inhibition	32, 34, 42, 75, 127, 129, 169, 171, 270
TAT-L2 peptide	Cx43 Hemichannel inhibition	279
Panx1 and Panx3 targeting peptides	Panx hemichannel inhibition	171a, 271
Apyrases	ATP degradation	39, 41, 42, 119, 129, 137, 169, 193, 236, 245, 257, 259, 261, 270, 295, 296, 300, 317, 358, 370, 407, 415
Purinergic receptor antagonists (suramin, PPADS, and others)	P2Y receptor inhibition	39, 41, 42, 119, 129, 137, 161, 192, 193, 236, 245, 257, 261, 270, 295, 296, 300, 317, 348, 358, 370, 371a
U73122	PLC inhibition	136, 144, 240, 385, 402, 408, 415
Heparin	IP3R inhibition	33, 246
2-Aminoethoxydiphenylborate (2-APB)	IP3R inhibition+	415
Thapsigargin	SERCA inhibition	33, 55, 61, 246, 385, 406, 415

References only refer to work demonstrating effects on Ca²⁺ waves; a selection of other references can be found in section VIIIA.

^*Several gap junction channel/hemichannel inhibitors also inhibit P₂X₇ receptors (347). ⁺2-APB also inhibits connexin channels (363).

Table 3.

Factors or conditions modulating intercellular Ca²⁺ waves

Molecule or Condition	Effect	Reference Nos.
CT-truncation of Cx43	Inhibition	190
Cx43G138R	Potentiation	371a
S-nitrosylation of Cx43 at C271	Potentiation	345
Low extracellular Ca2+ or Mg2+	Potentiation	341
Low extracellular Ca2+	Inhibition	240
Manganese	Inhibition	369
Ectonucleotidase inhibition with ARL-67156	Potentiation	129
Panx1 and P2X7	Potentiation	313, 347
Panx3	Potentiation	171a
Phorbol esters	Inhibition	219, 240
Thrombin	Inhibition	75, 76, 278
TAT-Cx43CT peptide (last 10 amino acids of Cx43)	Removes thrombin inhibition	279
α1-Adrenergic stimulation (hippocampal astrocytes)	Inhibition	240
β-Adrenergic stimulation (cardiomyocytes)	Potentiation	199
Serotonin	Inhibition	28
Glutamate	Potentiation	28
Stanniocalcin-1	Potentiation	27
TNF-α	Inhibition	382
IL-1β	Potentiation	176
Pax6 gene deficiency	Inhibition	207
High glucose	Inhibition	126, 131
Glucocorticoids	Potentiation	325
Anti-epileptic drugs	Inhibition	367
Oxygen-glucose deprivation followed by reperfusion	Potentiation	172
Ischemia in vivo	Potentiation	93
Aβ peptide	Potentiation	60
Alzheimer mice expressing mutant Aβ precursor protein and presenilin 1	Potentiation	191

The reported effect relates to the size or the repetition frequency of intercellular Ca²⁺ waves.

A. Inhibitors of Gap Junction Coupling

Importantly, many gap junction inhibitors do not only inhibit connexin hemichannels but also inhibit P₂X₇ receptors (347) and several other ion channels (29, 404); therefore, these substances may affect ATP release and paracrine communication of ICWs. Studies with gap junction inhibitors should also control for their influence on both gap junctions and hemichannels, but it is difficult to distinguish between these two configurations of connexin channels. Some gap junction inhibitors have contrasting effects on gap junctions versus hemichannels; for example, quinine stimulates hemichannel opening rather than closure (222, 395). Other agents with opposing effects on gap junctions and hemichannels include lipopolysaccharide (LPS), basic fibroblast growth factor (bFGF), proinflammatory cytokines, and arachidonic acid (85, 86, 254, 290). An additional caveat is that connexin channel inhibitors may also affect pannexin channels. For example, Cx46 hemichannels are blocked by 50–100 μM carbenoxolone, whereas Panx1 hemichannels are blocked by ∼10 μM carbenoxolone (45). In contrast, flufenamic acid blocks Cx46 hemichannels with an IC₅₀ of ∼30 μM, whereas Panx1 hemichannels are only affected by concentrations 10 times larger (45). The anion transporter inhibitor probenecid can inhibit Panx1 channels, without influencing connexin channels (324), but this substance has many other targets.

While antibodies have been used to block both gap junctions (16, 233) and hemichannels (65), greater experimental control has been sought with connexin mimetic peptides (77) to inhibit connexin channels. These peptides are synthesized to mimic specific sequences of the extracellular loops of connexin proteins. For example, the peptides Gap26 and Gap27 contain the amino acid sequences VCYD and SRPTEK, respectively, to mimic the conserved amino acid sequences of the first and second extracellular loop of the connexin protein (391). These peptides are thought to interact with as yet undefined sites within the extracellular loops of connexins of opposing hemichannels (best documented for Gap26; Ref. 213) and thereby prevent the docking of two hemichannels to form a gap junction channel (24, 106). Although the rapid effect of some connexin mimetic peptides (225) has led to the suggestion of an alternative mode of action, it is more likely that the rapid effects of Gap26 and Gap27 are mediated (as indicated by decreased dye uptake and ATP release) by the inhibition of hemichannel opening (40, 42, 86, 107, 108, 208). Electrophysiological studies of single Cx43 hemichannel currents also corroborate rapid channel closure within 4 min by Gap26 and Gap27 (Leybaert L. and Wang N., unpublished observations). Thus it appears that short (minutes) or long (hours) exposure to peptides (Gap26 or Gap27) can be used to inhibit only hemichannels or both hemichannels and gap junctions, respectively (82, 88).

Synthetic peptides corresponding to the L2 sequence on the cytoplasmic loop of Cx43 inhibit hemichannel opening (279) without inhibiting gap junctions (321), offering an interesting approach to dissect the roles of gap junctions and hemichannels in ICW propagation. These peptides interfere with the intramolecular interaction of the COOH-terminal tail of Cx43 with the cytoplasmic loop, thereby preventing hemichannel opening. Thus the L2 peptide partly inhibits ICW propagation because of hemichannel suppression without inhibiting gap junction channels (279). Synthetic mimetic peptides are also available to inhibit pannexin hemichannels; ¹⁰Panx1 is a peptide that mimics a sequence on the first extracellular loop of the Panx1 protein (271) and a peptide is also available to block Panx3 hemichannels (171a). The hemichannel inhibitory effect of these peptides may be based on mechanisms that are similar to Cx43 inhibition by Gap26, or Gap27.

The extracellular ionic composition has a significant influence on connexin and pannexin channel permeability. Trivalent ions (e.g., lanthanum, La³⁺) have been used to inhibit connexin hemichannels (7, 68, 315), but care is required because such ions also block Ca²⁺ channels (235, 407). Similarly, a reduction of the extracellular Ca²⁺ concentration can trigger connexin hemichannel opening (20, 46, 131, 342, 343, 405) without influencing gap junctions or pannexin channels. However, applying reduced extracellular Ca²⁺ conditions has many other effects: it reduces the driving force for Ca²⁺ entry, induces electrical field changes at the plasma membrane, and may deplete ER/SR Ca²⁺ stores if applied over a prolonged time period. Depending on the predominating mechanism, ICWs can be inhibited (240) or stimulated (341) by low extracellular Ca²⁺ conditions (TABLE 3). Acidosis is generally believed to inhibit gap junctions (334), but it may also have an opposite effect (356). Hemichannels are also inhibited by acidosis (304, 375) and opened with alkalosis (315). The effect of acidosis on pannexin channels is not yet established.

B. Inhibitors of Vesicular and Nonvesicular ATP Release

Several substances have been used to interfere with vesicular release of paracrine messengers mediating ICWs, including bafilomycin A, an inhibitor of the vesicular H⁺-ATPase and brefeldin A, an inhibitor of vesicular trafficking (39). The neurotoxins botulinum toxin and tetanus toxin can also be used (66, 163), but non-neuronal cells need long incubation times as they lack the transporters necessary for cellular uptake. The use of active fragments instead of holoenzymes, combined with adequate membrane permeabilization approaches to get these agents into cells, is preferable.

C. ATP Degrading Enzymes

ATP degrading apyrases are useful to explore ATP-based ICWs and have well-defined specificity for ATP and ADP depending on the enzyme subtype (236). Consequently, a cocktail of apyrase enzymes with both ATPase and ADPase activities has been used to efficiently remove ATP and prevent the accumulation of ADP (129). Endogenous ectonucleotidase activity can be inhibited by ARL-67156 (129). These enzymes are complementary tools to purinergic receptor antagonists to investigate the contribution of paracrine purinergic signaling in ICW propagation especially because the latter have poor selectivity (discussed below).

D. Purinergic Receptor Antagonists

The pharmacology of purinergic receptor antagonists is extensive (reviewed in Refs. 174, 251) but also characterized by a relative lack of selective receptor antagonists. Suramin, PPADS, and reactive blue-2 are broad-spectrum purinergic receptor inhibitors that have been frequently used to investigate the purinergic component of ICWs (TABLE 2). Alternatively, receptor desensitization by prolonged exposure to specific and/or nonhydrolyzable ATP analogs can be used (34, 236).

E. Inhibitors of Intracellular Ca²⁺ Signaling

Compounds interfering with intracellular Ca²⁺ signaling block intracellular Ca²⁺ waves and therefore may also be used to inhibit ICWs. Inhibition of PLC activity with U73122, IP₃R opening with heparin or 2-APB, and SERCA activity with thapsigargin have been consistently found to inhibit the propagation of ICWs (TABLE 2). Ca²⁺ buffers can be introduced into cells to reduce Ca²⁺ diffusion. These effects have supported the view that ICWs largely rely on the IP₃-dependent Ca²⁺ release. In contrast, inhibition of RyR opening with dantrolene or high ryanodine concentrations reduced the magnitude of the [Ca²⁺]_i increase but had no effect on ICW propagation (55, 144, 385, 408). This result emphasizes the idea that substantial [Ca²⁺]_i increases are not vital for ICW propagation.

The alternative approach of using compounds that stimulate, rather than inhibit, ICWs is equally powerful and perhaps is best illustrated by the microinjection of IP₃ or photolytic release of caged IP₃. A variety of other caged compounds such as caged Ca²⁺, Ca²⁺ ionophores, ATP, and glutamate are available. A particular advantage of photolysis techniques is that they are relatively noninvasive and highly compatible with imaging approaches being used for in vivo studies.

IX. FUNCTIONS OF INTERCELLULAR Ca²⁺ WAVES

In principle, ICWs mediate the transmission of information from a local site to a global area; the communication of Ca²⁺ signals to multiple surrounding cells provides the potential to coordinate and synchronize the function of a large group of cells. Cell activities known to be stimulated by increases in [Ca²⁺]_i, such as PKC activation (288) or vesicle secretion, would be expected to be enhanced in cells participating in the ICW. A good example of this correlation between cell activity and an ICW is observed in airway ciliated epithelial cells, where the passage of an ICW initiated an increase in ciliary activity (202). Because cilia are more effective at moving mucus when beating in coordinated groups, this behavior supports the idea that ICWs serve to coordinate tissue responses. Numerous other coordinating functions for ICWs have been proposed; these include proliferation or differentiation of keratinocytes (379), upstream signaling in the blood vessel wall (180, 362), signaling between smooth muscle cells (141), and signaling between endothelial cells and smooth muscle cells via myo-endothelial gap junctions to control vascular tone (170), communication between various cells in the juxtaglomerular apparatus mediating tubuloglomerular feedback (333, 403), and regulation of metabolic activity of glial cells (21, 258). ICWs have also been proposed to provide communication between different cell types, for example, glial-neuron interactions in the brain, endothelial-glial interactions at the blood-brain barrier, and endothelial-smooth muscle cell communication in the vasculature (TABLE 1).

One concern regarding the function of ICWs, either in vitro or in vivo, is whether they result from strong experimental stimulation that does not have a normal physiological counterpart. On the other hand, certain pathological conditions, such as trauma, are often associated with strong stimuli. Brain trauma, brain ischemia, and seizures are all coupled with the release of ATP and glutamate and a decrease of extracellular Ca²⁺ (142, 149, 209) that collectively promote ICW generation. The excessive elevation of [Ca²⁺]_i that is likely to be associated with these stimuli can lead to Ca²⁺-dependent PLC activation, the generation of large amounts of IP₃, and the initiation of ICWs. Such ICWs may exacerbate injury by communicating signals that initiate cell apoptosis in surrounding cells (90, 196, 197, 398). Recent evidence indicates that IP₃ diffusion via gap junctions is crucial to provoke apoptosis in adjacent cells (89). Conversely, ICWs may provide the directional clues to the injury site for subsequent recruitment of repair mechanisms. This correlation of ICWs and tissue repair has also been proposed for hepatic tissue (355) and vascular endothelial cells (in concert with serum growth factors) (373). In line with this, decreased ICW activity may delay wound healing (207).

A second unknown with respect to ICWs in physiological or pathological responses is the signaling molecule(s) that mediates the response. This could be the resulting [Ca²⁺]_i changes, the messengers propagating the ICWs (i.e., IP₃ or ATP), or another associated signal. For example, ICWs that are proposed to coordinate metabolic activity of glial cells (54) probably mediate their effect by the changes in [Na⁺]_i that accompany the ICW rather than by changes in [Ca²⁺]_i (21, 200, 258). ATP is a likely candidate to mediate pathological responses to tissue trauma by exerting a mitogenic action or acting as a “find me” signal from apoptotic cells for phagocytic clearance (101). ATP also triggers directed cellular responses in brain injury, such as microglial activation and chemotaxis (80), and has a wide range of targets in the brain and immune system (48, 92). Its metabolite adenosine may also confer a protective action in the brain.

A further caution regarding the roles for ICWs is that most hypotheses emanate from observations of in vitro cell culture systems. However, many aspects of ICW initiation and propagation may be very different under in vivo conditions. For example, in organs, the diffusion of intra- and extracellular messengers can occur in three rather than the two dimensions typical of monolayer cultures. This spatial distribution is likely to influence the distance of ICW travel as well as the clearance or metabolism of extracellular messengers by the microcirculation or ectonucleotidases (97). Equally important is the fact that in tissue culture, many of the organ characteristics, such as the extracellular matrix, variety of cell types, extent of cell coupling, and cell differentiation, are lost or significantly altered. At present, there are fewer reports of ICWs in vivo in organs like the brain, liver, and lung (161, 192, 265, 292) and therefore, as yet, insufficient supportive data for specific physiological responses of ICWs. However, this situation is rapidly changing. Below, we discuss several examples where ICW propagation and function were studied in intact tissues in vitro or in animals in vivo (FIGURE 4).

Figure 4.

Overview of mechanisms and functions of intercellular Ca²⁺ waves (ICWs), based on ex vivo and in vivo evidence from the organs indicated. ICWs between cells of the same type are indicated by circular arrows; noncircular arrows indicate ICWs between different cell types. Gap junctions (GJs) and paracrine purinergic signaling (PPS) are involved in most cases, but only the most prevalent mechanism is depicted. Dashed arrows indicate proposed functions (in italics). Retina: ICWs that propagate in astrocytes and Müller cells may provoke electrical activity in retinal ganglion cells (RGCs) and influence the blood vessel caliber. Retinal pigment epithelium cells (RPE) show ICWs which may influence proliferation and differentiation of neural progenitor cells (NPCs). Cochlea: ICWs in supporting cells may be involved in K⁺ recycling and may influence outer hair cells (OHCs) via ATP release to alter the cochlear amplifier gain and inner hair cells (IHCs) to modify electrical activity and synaptic connectivity during development. Blood vessels: ICWs propagate along the vessel wall via endothelial cells (ECs) and modulate white blood cell (WBC) interactions with ECs as well as smooth muscle cell (SMC) contractility via direct (myoendothelial gap junctions) or indirect mechanism to control vessel diameter. Brain: mechanisms of ICW propagation differ with brain region; GJs, GJs+PPS, or PPS mediate ICWs in the neocortex, hippocampus, and corpus callosum and Bergman glia, respectively. ICWs of astrocytes propagate via astrocyte endfeet to small-diameter blood vessels to influence blood vessel diameter. ICWs of astrocytes may also modulate synaptic signaling as well as contribute to neural development or disease processes (AD, Alzheimer disease; CSD, cortical spreading depression). Liver: ICWs propagating between hepatocytes may influence hepatic glucose output, bile production, and tissue regeneration.

A. Intercellular Ca²⁺ Waves in the Retina

Stimulated ICWs occur in astrocytes and Müller cells of isolated retina (rat eyecup, whole-mount retina, and retinal slices) (245, 246) and influence the electrical activity of retinal ganglion cells (the output neurons of the retina), suggesting that ICWs are involved in glial-neuronal signaling (247). ICWs between astrocytes are dependent on gap junctions, whereas ICWs between Müller cells are more dependent on paracrine purinergic signaling, although they utilize some gap junction signaling. ATP release via connexin hemichannels has been demonstrated in Müller cells (44). ICWs also propagated between astrocytes and Müller cells; although these astrocytes and Müller cells are coupled by rectifying gap junctions (246, 411), this heterocellular communication of ICWs appeared to be mediated by paracrine signaling rather than gap junctions.

Spontaneous ICWs occur in Müller cells but not in astrocytes in the retina during postnatal development. The frequency of these spontaneous ICWs increased with age (193), both in vitro and in vivo (open eye globe), and this may result from increased ATP release and enhanced ATP regeneration. When ICWs in the in vitro retina encountered a blood vessel, they induced vasoconstriction, a response indicating a role for glial ICWs in controlling blood flow. It remains to be determined if spontaneous ICWs influence the electrical activity of retinal ganglion cells, as observed for stimulated ICWs.

ICWs may also contribute to the spontaneous electrical and Ca²⁺ wave activity in the ventricular zone of the retina (357) and retina proper (51) during retinal development. These ICWs are mainly restricted to neurons (although glia have not been excluded) and are important to guide development of the synaptic connectivity in utero in the absence of visual input (181).

Spontaneous ICWs propagate repeatedly (several waves per minute) between retinal pigment epithelial (RPE) cells and the neural retina. These ICWs are mediated by extracellular ATP that is released from trigger cells via hemichannel opening, based on evidence of propidium iodide dye uptake and inhibition by Gap26 (270). The mitogenic action of the ATP associated with these waves is implicated in controlling proliferation and differentiation of retinal neural progenitor cells (269).

B. Intercellular Ca²⁺ Waves in the Cochlea

In mouse cochlear explant cultures (postnatal day 3 to 6), focal stimulation of supporting cells, located in the organ of Corti, with extracellular ATP or by photolysis of IP₃ triggered ICWs propagating in the supporting cell zone (7) (FIGURE 4). The involvement of ATP as an extracellular messenger released via connexin hemichannels and gap junctions was confirmed with studies with Cx26- and Cx30-deficient animals, connexin channel blockers (including La³⁺ to block hemichannels), and ATP biosensors (7, 221). Panx1 hemichannels and P₂X₇ receptors were not involved. ATP-based ICWs in supporting cells are known to trigger Ca²⁺ spikes in inner hair cells and electrical activity in spiral ganglion cells (377). This electrical activity may help to establish and organize synaptic connectivity within the primary auditory cortex (A1) before auditory input occurs.

Whether cochlear ICWs contribute to the hearing in the adult is not known. Because mutations in both Cx26 and Cx30 genes are frequently associated with autosomal recessive nonsyndromic hearing loss (10), it is hypothesized that gap junction dysfunction impairs K⁺ clearance from the inner hair cells following auditory stimulation (414). This concept is underscored by the deafness-associated mutation where a V84L substitution in Cx26 selectively impairs the permeability of Cx26 gap junctions to IP₃ and thereby strongly reduces ICW propagation (17). ICWs may be involved in K⁺ recycling via supporting cells by activating K⁺ efflux from these cells (223). Extracellular ATP (at micromolar concentrations) decreases the electromotility of the outer hair cells that function as an active amplifier in the organ of Corti (417). Consequently, an ATP wave associated with the ICW is expected to influence the hearing sensitivity.

C. Intercellular Ca²⁺ Waves in the Brain

1. In vitro tissues

ICWs have been observed in vitro in acutely isolated (ex vivo) brain tissues (250). Astrocytes within brain slices from the rat thalamus display spontaneous Ca²⁺ oscillations that propagate to a limited number of surrounding astrocytes and trigger inward currents in associated neurons (264). Local electrical stimulation of mouse cortical brain slices triggers extensive (several 100 μm) fast propagating (∼50 μm/s) astrocytic ICWs. The fast propagation reflects a contribution by the neuronal network because tetrodotoxin (blocks action potentials) or Ca²⁺-free solutions (blocks presynaptic neurotransmitter release) significantly reduced wave speed (139).

However, the mechanism of ICW propagation appears to vary with brain location (FIGURE 4). In astrocytes of the neocortex, studies with dye coupling, gap junction inhibitors, and Cx43 knockout (astrocytes) animals (139) indicated that gap junction coupling was extensive and had a dominant role in ICW communication. Although the same astrocytes express Cx30, this was secondary to Cx43 in ICW propagation. Some ATP release (detected by “sniffer cells” expressing metabotropic purinergic receptors) was associated with neocortex ICWs, but these were not inhibited by purinergic receptor antagonists (139), suggesting ATP was not important. In compact tissues like the brain, ATP is probably rapidly degraded by ectonucleotidases, and this precludes the activation of astrocytic purinergic receptors.

In the corpus callosum, electrical stimulation triggers glial ICWs that propagate into the neocortex. In contrast, the same electrical stimulation in Cx30/43-deficient mice initiated ICWs that failed to propagate into the neocortex. This indicates that ICWs within the corpus callosum, unlike the neocortex, do not utilize gap junctions. Similarly, ICWs were still present in the CA1 region of the hippocampus of Cx30/43-deficient mice. These ICWs were inhibited by purinergic antagonists but not a gap junction blocker, indicating a purinergic rather than gap junction mode of propagation (139). However, recent evidence indicates that ICWs triggered in hippocampal slices by diazo-2 photoactivation are strongly inhibited in Cx30/43-deficient mice (371a).

2. In vivo ICWs

In vivo studies of ICWs can be achieved by two-photon microscopy; this technique allows the noninvasive monitoring of [Ca²⁺]_i changes up to a depth of ∼400 μm (252), which corresponds to cortical layers II and III of mice and rats. The loading technique used to introduce Ca²⁺ indicator dyes into cells may affect the results; glial cells, more specifically astrocytes, are predominately loaded with indicator dye by applying the dye to the cortical surface, whereas both glial and neuronal cells are loaded by a bolus injection of dye into the cortex (120, 340, 354, 368). Alternatively, cells may express a genetically encoded GCaMP2 Ca²⁺ indicator (161) to avoid mechanical stimulation of the neural tissue associated with dye injection. Although two-photon microscopy is noninvasive, there can be some artifacts associated with two-photon laser excitation. For example, excitation with high-energy pulses of infrared light may locally generate free oxygen radicals (due to triple-state excitation) or thermal effects, and these may serve as triggers to stimulate ICWs (389). Conversely, too much laser energy may dampen Ca²⁺ wave activity (192).

Similar to the observations in brain slices, extensive astrocytic ICWs (propagating at 8–10 μm/s over 150–200 μm) were triggered by the photolytic release of Ca²⁺ in a single astrocyte or by local iontophoretic application of ATP in the cortex of anesthetized rodents (368). This initiation of ICWs by the photolytic release of Ca²⁺ is in stark contrast to the response commonly observed in cell culture where a Ca²⁺ increase alone does not trigger an ICW (except in special conditions as discussed). This suggests that the experimental elevation of [Ca²⁺]_i in vivo may stimulate a larger increase in IP₃ or ATP release in the target cell. Alternatively, Ca²⁺ may serve as a better intercellular messenger in vivo because of a preexisting elevation of [IP₃]_i in adjacent cells or because gap junction coupling is more extensive. In addition, in vivo experiments are conducted at body temperature, and this may increase the Ca²⁺ excitability of the tissue.

Spontaneous ICWs occur in cerebellar cortex of rodents (rats or mice, 3–6 wk old) (161). These ICWs propagated in Bergmann glial cells in three dimensions (radius of ∼50 μm at 4–11 μm/s) with a frequency of ∼0.4 waves/min. The major axes of propagation were aligned in one direction with the parallel fibers and in another direction with the pia-Purkinje cell axis and corresponded to the directions in which extracellular diffusion occurred at its faster rate. This behavior is consistent with extracellular purinergic signaling; gap junction coupling was presumed to be inhibited by the [Ca²⁺]_i elevation associated with the ICW (161). Similar spontaneously occurring ICWs (called “Ca²⁺ bursts”) have been described by Nimmerjahn and co-workers at about the same time (250a); importantly, the latter work was performed in the absence of anesthesia.

Coordinated [Ca²⁺]_i changes (not “true” waves) also occur in cortical astrocytes in rats (P12-P16) after stimulation of neuronal activity with bicuculline, a suppressor of GABAergic neurotransmission (154). This suggests spontaneous astrocytic wavelike activity can result from increased neuronal network activity. Spontaneous ICWs called “glissandi” were found in the hippocampus of mice (192) and encompassed several tens of cells, presumably astrocytes, and propagated approximately every 3 min at a high speed (∼61 μm/s). Glissandi were inhibited by carbenoxolone and suramin, indicating gap junctions and ATP signaling in their propagation, and by tetrodotoxin, suggesting a neuronal network contribution.

Most of the two-photon studies detailed above were performed in anesthetized animals. However, an optic fiber positioned on the cortex has been used to observe [Ca²⁺]_i changes in active, nonanesthetized animals. Interestingly, only limited Ca²⁺ wavelike activity, presumed to be of neuronal origin, was observed (3). Importantly, this Ca²⁺ wave activity disappeared under anesthesia. This implies that spontaneous ICW activity recorded with two-photon imaging in anesthetized animals may become more pronounced if the anesthesia could be omitted. Work of Nimmerjahn et al. (2009) performed in behaving mice has shown that the ‘Ca²⁺ burst’ type of ICWs disappears with the start of locomotive behavior; instead, so called ‘Ca²⁺ flares’ appear that involve concerted Ca²⁺ excitation in hundreds of Bergman glial cells (250a). However, Ca²⁺ flares are not ICWs.

3. Proposed functions

A) SYNAPTIC MODULATION.

Because astrocytes make substantial contacts with neurons, it is speculated that astrocytic ICWs influence synaptic activity (FIGURE 4). Astrocyte extensions are closely apposed to neurons in tripartite synapses (274), and increases in astrocytic [Ca²⁺]_i as well as ICWs have been shown, in vitro, to influence synaptic functioning by glutamate release (113, 273), although this Ca²⁺ dependency of gliotransmitter release is not universal (4). Similarly, in vivo glissandi waves were associated with a reduction of local field potentials [representing dendritic electrical activity (192)]. Although neuronal activity is known to trigger localized astrocytic [Ca²⁺]_i changes (319, 389), there is currently no in vivo evidence that neuronal activity can trigger astrocytic ICWs. However, in vitro, astrocytic ICWs may be triggered by extracellular factors, such as pH changes; brain stem slices display astrocytic ICWs in response to a lowering of the extracellular pH (133).

B) BLOOD FLOW REGULATION.

Astrocytes also form endfeet extensions contacting small-diameter brain arteriolar blood vessels, and [Ca²⁺]_i changes in these endfeet influence vessel diameter and blood flow (12) (FIGURE 4). The vessel response to astrocyte Ca²⁺ stimulation appears to vary between different recording conditions. In brain slices, photolytic release of Ca²⁺ in astrocytic endfeet induced vasoconstriction while vasodilation was observed in vivo in the somatosensory cortex (239, 360). In brain slices, ICWs occurred along the blood vessels [similar findings occurred following electrical stimulation (115)] while ICWs were absent in vivo. In contrast, spontaneous in vivo hippocampal glissandi waves were associated with decreased blood flow in their propagation territory (192). A similar dual vasoconstriction or vasodilation to astrocyte stimulation occurs in the retina (232). In the brain, a low oxygen tension appears to favor vasodilation (132).

Astrocytic ICWs have been proposed to play a role in the spatial buffering of K⁺, with K⁺ being released at endfeet via large-conductance Ca²⁺-sensitive K⁺ channels. In arterioles, K⁺ can induce both vasodilation and vasoconstriction depending on the concentration of K⁺ in the extracellular space (116, 124). Low K⁺ concentrations hyperpolarize vascular smooth muscles to induce relaxation, whereas high K⁺ concentrations depolarize smooth muscle to induce contraction. However, the contribution of astrocyte-derived K⁺ to vascular responses remains contentious (231).

C) DEVELOPMENTAL REGULATION.

Glial ICWs may play a role in the organization and proliferation of the ventricular zone of the developing neocortex (observed in brain slices) (394). Radial glial cells display spontaneous and repetitive ICWs that primarily utilize hemichannel opening to release ATP and P₂Y₁ receptor signaling, although some gap junction communication was evident. Like the retina, ATP is proposed to act as a mitogenic signal to synchronize cell cycles, thereby acting as a spatial signal coordinating the activity of radial glial cells within the ventricular zone (214). The role of the [Ca²⁺]_i changes associated with ICWs remains to be determined (79).

D) PATHOPHYSIOLOGY.

Astrocytic ICWs also appear under pathological conditions (FIGURE 4). For example, spontaneous ICWs occur near amyloid-β plaques in adult APP/PS1 Alzheimer mice (191). Evidence for wave activity (or rather coordinated [Ca²⁺]_i changes) comes from cross-correlogram studies of multiple cells and indicates a propagation speed of ∼23 μm/s and travel distances of 200 μm from cortical plaques. This coordinated activity was not influenced by tetrodotoxin and amyloid-β and was speculated to trigger the activity via its neurotoxic effects. The propagation mechanisms of the coordinated [Ca²⁺]_i changes (studied in vitro) suggest dependence on both gap junctions and paracrine ATP signaling (60, 148, 219, 241). Astrocytic coordinated [Ca²⁺]_i changes occur 20 min after photothrombotic obstruction of a small vessel (0.6 mm in diameter) in the brain cortex of mice (93). In this case, the coordinated [Ca²⁺]_i changes were inhibited by antagonizing type 5 metabotropic glutamate receptors or GABA-B receptors; purinergic or adenosine receptor antagonists had no influence. Interestingly, the loading of astrocytes in vivo with a Ca²⁺ buffer (BAPTA-AM) reduced the infarct size, pointing to a possible role of Ca²⁺ wavelike activity in ischemic cell damage.

ICWs have been implicated in cortical spreading depression (CSD) which may contribute to the pathophysiology of the aura phase of migraine (52, 359). CSD represents a wave of cortical neuronal depolarization. Experimentally, CSD can be triggered mechanically or by the local application of a high K⁺ concentration to the cortex. CSD (observed by intrinsic optical signal imaging in brain slices) propagates quickly with a high extracellular K⁺ concentration at the leading edge, followed by depressed cortical activity (reviewed in Ref. 359); this CSD wave is associated with a slower propagating astrocytic ICW (61). The CSD wave and the associated ICW have distinct mechanisms: CSD is inhibited by NMDA receptor antagonists, while the ICW is inhibited by gap junction inhibitors, indicating a neuronal and astrocytic contribution, respectively (275). The inhibition of the astrocytic ICW by thapsigargin correlated with a loss of early vasoconstriction associated with CSD. This implies a role for astrocytic ICWs in the acute hypoperfusion phase of CSD. However, ICWs probably do not play a role in the hyperperfusion and long-lasting hypoperfusion that follow the acute hypoperfusion phase (61).

D. Intercellular Ca²⁺ Waves in Blood Vessels

While it is clear from the previous discussion that ICWs propagating through the brain parenchyma may influence vascular tone, ICWs also propagate relatively long distances along the cells of blood vessels to modulate vascular function (FIGURE 4). In venous lung capillaries of isolated, blood-perfused lungs from rodents, the photolytic release of Ca²⁺ in an alveolar endothelial cell triggered ICWs propagating from the capillaries to venous blood vessels, up to 150 μm away from the stimulation point (265). The mechanisms by which Ca²⁺ triggers ICWs were addressed previously (see sect. IXC). These ICWs were inhibited by Gap26 and Gap27 and were absent in mouse lungs with an endothelium lacking Cx43, results consistent with gap junction communication. Interestingly, the presence of ICWs correlated with increased endothelial P-selectin expression and potentiation of thrombin-induced microvascular permeability. Because both of these responses were also attenuated by Gap26/Gap27, it is feasible that these cellular changes were induced by the ICWs. Lung capillary endothelial ICWs also occur spontaneously, at a frequency of ∼ 0.4 waves/min (406). Long-range ICWs, propagating several hundred micrometers, have been reported in intact isolated blood vessels from heart and cremaster muscle (362). Here, the ICWs were shown to induce vasodilation and thereby contribute to the regulation of tissue perfusion.

E. Intercellular Ca²⁺ Waves in Liver

Confocal imaging of the superficial cell layers of intact perfused liver has demonstrated ICWs within the hepatocyte plates in response to the hormones vasopressin, phenylephrine, or angiotensin II (238, 242, 268, 292). However, as discussed earlier, the propagation of liver ICWs does not rely on a messenger diffusion-reaction cascade but appears best represented by a phase wave where hormones trigger time-delayed Ca²⁺ oscillations with a similar frequency in adjacent hepatocytes; the asynchronous, but sequential Ca²⁺ increases give the impression of an ICW.

Hepatocyte ICWs in intact liver appear to be initiated from periportal areas and propagate in the direction of the central vein (122), but initiation from the pericentral vein can also occur (98, 238). The waves are limited to the microcirculatory domain of a hepatic lobule and do not spread to neighboring microcirculatory domains. Liver ICW activity is inhibited by gap junction closure or by disruption of cell-cell contacts, but the hormone-induced Ca²⁺ oscillations persist, albeit in a less coordinated fashion (122). There is little evidence for a contribution of purinergic paracrine communication (268). Mathematical modeling predicts that the direction of ICW propagation is determined by a gradient of hepatocyte hormone sensitivity (and receptor densities) along the hepatic plate, while the gap junction coupling appears to be essential for the entrainment or establishing the phase delay in Ca²⁺ spiking between cells (99, 100). Gap junctions appear to reduce the [IP₃]_i gradient between cells that is initially established within hepatocyte cells upon exposure to hormone.

Hepatocyte ICWs are hypothesized to play a role in hepatic glucose output, activation of bile flow, and liver regeneration after partial hepatectomy (FIGURE 4). Hepatic glucose output triggered by norepinephrine or glucagon is significantly lower in Cx32-deficient animals compared with wild type (346). Given the importance of cAMP and IP₃ in hepatic glucose mobilization, it is plausible that the diffusion of these two messengers via gap junctions helps recruit hepatocytes to provide sufficient glucose output. However, a clear demonstration of the role of the ICW is still lacking. Vasopressin is normally a strong trigger of bile secretion, but after 24 h of in vivo exposure to elevated vasopressin concentrations, a pulse of vasopressin triggered significantly less bile flow. This treatment removed the hepatocyte vasopressin receptor gradient in the perivenous area and impaired the ICWs, suggesting a link between the ICWs and bile flow (322). Adenoviral transfection of hepatocytes with the Ca²⁺-buffering protein parvalbumin reduced liver regeneration after partial hepatectomy and reduced norepinephrine-induced Ca²⁺ oscillations, pointing to the importance of hepatocyte Ca²⁺ signaling in liver regeneration (195). However, it is yet unclear whether parvalbumin transfection has any influence on hepatocyte ICWs.

X. CONCLUSIONS

ICWs appear to be a widespread phenomenon by which a diversity of cell types communicate with each other to coordinate and synchronize their activity. While ICWs occur both in vitro and in vivo, there are some differences in behavior and mechanisms that indicate close attention should be paid to experimental conditions and tissue characteristics to accurately identify their physiological roles. ICWs appear to utilize two fundamental mechanisms of propagation that may either operate in isolation or in combination. These are the diffusion of IP₃ through gap junctions and the release of extracellular ATP that serve as messengers to stimulate adjacent cells via the direct release of Ca²⁺ from the ER or the indirect release of Ca²⁺ by stimulation via purinergic receptors, respectively. The stimuli initiating ICWs are also varied, but the fact that ICWs can spontaneously occur in vivo points to their existence in normal physiological conditions. ICWs also appear to be involved in pathological processes such as brain ischemia and perhaps Alzheimer's disease. While the identity of the actual messenger associated with the ICW that is responsible for such control is uncertain, ATP is most commonly implicated. Future studies aimed at elucidating the extent and function of ICWs in vivo promise to reveal important insights into a wide diversity of processes including neural development, hearing, vision, liver function, and brain disease.

GRANTS

L. Leybaert was supported by the Fund for Scientific Research, Flanders, Belgium (G.0140.08, 3G.0134.09, G.0298.11N, and 3G057112N), and the Interuniversity Attraction Poles Program, Belgian Science Policy (P6/31 and P7). M. Sanderson was supported by National Heart, Lung, and Blood Institute Grant HL-103405.

DISCLOSURES

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