IP₃ receptor signaling and endothelial barrier function

Abstract

The endothelium, a monolayer of endothelial cells lining vessel walls, maintains tissue-fluid homeostasis by restricting the passage of the plasma proteins and blood cells into the interstitium. The ion Ca²⁺, a ubiquitous secondary messenger, initiates signal transduction events in endothelial cells that is critical to control of vascular tone and endothelial permeability. The ion Ca²⁺ is stored inside the intracellular organelles and released into the cytosol in response to environmental cues. The inositol 1,4,5-trisphosphate (IP₃) messenger facilitates Ca²⁺ release through IP₃ receptors which are Ca²⁺-selective intracellular channels located within the membrane of the endoplasmic reticulum. Binding of IP₃ to the IP₃Rs initiates assembly of IP₃R clusters, a key event responsible for amplification of Ca²⁺ signals in endothelial cells. This review discusses emerging concepts related to architecture and dynamics of IP₃R clusters, and their specific role in propagation of Ca²⁺ signals in endothelial cells.

Introduction

In the late 19th century, Wilhelm His Sr. first introduced the term “endothelium” when he described differences between the layers of cells that line the mesoderm. After the inception of the idea of an endothelium as a specific tissue type or organ, the understanding of its exact location spanned many years of debate and technological advancements (reviewed in [1]). Notedly, with the discovery of its form, came the discoveries of its function. In the recent century, we have transitioned from asking, where the endothelium exists to determining the precise function of the endothelium and how this function is regulated in health and disease.

The endothelium is considered as an autocrine, paracrine, and endocrine organ capable of regulating a broad range of vascular functions including tissue-fluid homeostasis, vascular tone, thrombogenesis, inflammation, and vessel growth ([2–4]; reviewed in [5–7]). The large body of data related to the functions of endothelial cells has nowadays centered upon an important feature: vascular permeability. Considering that the inner vascular lining is formed by mature, quiescent endothelial cells, its primary function is not to proliferate or migrate but rather to act as a semi-permeable barrier between the vascular system and surrounding tissues.

Permeability of the endothelial barrier represents one of the hallmark processes for the maintenance of proper tissue-fluid homeostasis [6–8]. Once the endothelial barrier is compromised in response to humoral or mechanical stimuli, the barrier loses its selective control, causing excessive leakage of protein-rich fluids across the barrier. Vascular leakage is not only an integral part of many diseases—including systemic capillary leak syndrome [9], dengue fever [10] and Ebola viruses [11, 12], angioedema [13], anaphylaxis [14], acute lung injury [15, 16], variety of eye [17], and central nervous system [18, 19] disorders—but also the major dose-limiting factor in many immunotherapies [20]. The basis of such an event lies in the disruption of the inter-endothelial junctions, which are finely regulated, in part, by intracellular calcium signaling.

The ion Ca²⁺ is a secondary messenger paramount to a multitude of physiological and pathological processes in human body—muscle contraction, metabolic regulation, and fertilization to name a few. Within endothelial cells, the ion Ca²⁺ also plays a fundamental role in the majority of intracellular processes including, but not limited to, cell proliferation, motility, cell–cell adhesion, and cell death. While few recent reviews focus on the role of intracellular calcium in regulation of vascular tone through control of nitric oxide production [21–23], endothelial cell migration and proliferation—a hallmark of angiogenic response of mature endothelium [24, 25] and endothelial progenitor cells [26] to environmental cues, we focus here on the role of IP₃R channels, an ion channel whose gating is important to the control of calcium homeostasis. Novel findings in IP₃R architecture and dynamics give further insight into regulation of IP₃-evoked calcium signaling and, thereby, regulation of endothelial permeability.

The role of intracellular Ca²⁺ signaling in regulating endothelial barrier function

Endothelial barrier function

Endothelial cells form an inner monolayer, termed the endothelium, lining the vessels of the blood and lymphatic systems, and the endocardium. The endothelium acts as a semi-permeable barrier that regulates passage of fluids, solutes, gases, nutrients, as well as transmigration of blood cells from circulation into surrounding tissues. The passage of macromolecules such as albumin across the endothelial barrier is essential to the maintenance of tissue-fluid homeostasis. Two main pathways, paracellular and transcellular, are involved in this process. The transcellular pathway, also known as transcytosis, uses receptor-mediated vesicular transport to actively distribute macromolecules across the endothelium. In contrast, the paracellular pathway utilizes interendothelial cell–cell junctions that connect adjacent endothelial cells and restrict permeability to macromolecules larger than 3 nm in diameter [27–29]. Hence, albumin, a molecule of 3.8 nm in diameter and 15 nm in length [30, 31], which represents ~60% of all plasma proteins [32, 33], is mainly retained in circulation. This higher concentration of albumin in the circulation compared to the interstitial space creates transendothelial oncotic pressure, a driving force of fluid reabsorption from the interstitium [34, 35]. Loss of interendothelial junctions in disease settings causes proteinaceous tissue edema, a pathological condition associated with leakage of protein-rich fluids in the interstitum [13, 15, 16].

Interendothelial junctions are comprised of adherens junctions (AJs), tight junctions (TJs), and gap junction (GJs) [36–39]. AJs, composed of Vascular Endothelial (VE)–cadherin adhesion complexes, are the primary cell–cell adhesions in the endothelium except for the brain and retinal blood barriers [40–42], also reviewed in [7]. Numerous signaling pathways—including those induced by free cytosolic calcium (Ca²⁺)—regulate the integrity of AJs [43–47]. The ion Ca²⁺ is a versatile and ubiquitous secondary messenger implicated in destabilization of endothelial barrier function [48–53]. It participates in signal transduction by binding to intracellular proteins, most of which contain the EF-hand motif [54, 55], and causing conformational changes in their tertiary structure [56, 57]. In most cases, these conformational changes result in activation of protein functions.

The role of calcium-dependent kinases

In endothelial cells, the ion Ca²⁺ orchestrates a set of signal transduction events through activation of calcium-dependent kinases and phosphatases (Fig. 1). These signaling molecules alter endothelial barrier function by multiple convergent mechanisms [6, 7]. The ion Ca²⁺ activates serine/threonine kinases including protein kinase Cα (PKCα), myosin light chain kinase (MLCK)-210 (endothelial-specific isoform), and Ca²⁺/calmodulin dependent protein kinase II (CaMKII). In turn, these kinases induce disassembly of AJs, thereby increasing permeability of endothelial barrier to plasma proteins [47, 58–64]. For example, PKCα phosphorylates p120-catenin [47, 65, 66], a protein of the VE–cadherin adhesion complex that binds to the juxtamembrane region of the VE–cadherin [67]. Phosphorylation of p120-catenin at the S879 residue decreases binding affinity of p120-catenin to VE–cadherin, resulting in dissociation of p120-catenin from the VE–cadherin adhesion complex [47]. This event enables binding of an adaptor protein (AP)-2 complex to VE–cadherin [68, 69] and subsequent initiation of internalization [47, 67] and proteolytic processing of VE–cadherin [70, 71].

Fig. 1.

Intracellular Ca²⁺ signaling in endothelial cells associated with vascular leakage and inflammation. Increase in cytosolic Ca²⁺ concentration activates specific kinases and phosphatases that, in turn, promote vascular leakage and inflammation. The serine/threonine kinase PKCα induces phosphorylation of p120-catenin and, through this mechanism, contributes to disassembly of AJs. PKCα also promotes reorganization of actin cytoskeleton and acto-myosin contractility by activating RhoA signaling. CaM-dependent activation of both MLCK-210 and CaMKII facilitates reorganization of actin cytoskeleton through the phosphorylation of actin motor myosin-II and caldesmon. In addition, CaM activates the phosphatase CaN, which dephosphorylates the microtubule accessory factor EB3, thereby coordinating reorganization of the actin and microtubule cytoskeleton. CaN also mediates activation of NFAT and promotes cytokine production by endothelium

PKCα is also responsible for the activation of RhoA, a small GTPase that reorganizes the actin cytoskeleton through activation of the downstream effectors, Rho-associated coiled-coil forming protein kinase (ROCK) and Diaphanous-related formin-1 (mDia1) [72–74]. PKCα induces activation of RhoA signaling by phosphorylating both a Guanine Nucleotide Dissociation Inhibitor (GDI) [44, 75] and a Guanine Nucleotide Exchange Factor (GEF) p115RhoGEF [76–78]. Phosphorylation enables p115RhoGEF-mediated GTP exchange by inhibiting GDI binding to RhoA [75, 76]. RhoA activates ROCK, which in turn phosphorylates myosin light chain phosphatase (MLCP) at Thr-695, Ser-894, and Thr-850 [79–81], inhibiting MLCP activity [79]. This process enables phosphorylation of myosin-II, an actin motor that assembles the acto-myosin contractile apparatus, at Ser19/Thr18 by MLCK-210 [82–84]. RhoA also mediates translocation of mDia1 to AJs, allowing polymerization of F-actin filaments [85]. mDia1-dependent assembly of actin filaments is reinforced by CaMKII-dependent activation of Mitogen-activated protein kinase 3/1 (MAPK3/1 or ERK1/2), which phosphorylates actin-binding protein caldesmon at Ser789 to prompt re-organization of the actin cytoskeleton [58, 60, 63].

Another mechanism of endothelial barrier destabilization involves proline-rich tyrosine kinase 2 (Pyk2) that is responsible for phosphorylation of Vascular Endothelial Tyrosine–Protein Phosphatase (VE–PTP), a constituent of AJs [86, 87]. VE–PTP, also known as PTPβ, provides a constitutive mechanism for VE–cadherin dephosphorylation at critical tyrosine residues Y658 and Y685 and, hence, prevents VE–cadherin internalization [88, 89]. Phosphorylation of VE–PTP at Y1981 residue causes a dissociation of VE–PTP from VE–cadherin, thereby increasing endothelial permeability [89]. In addition, Pyk-2 positively regulates Ca²⁺ entry in endothelial cells by inducing phosphorylation at Y361 of stromal interaction molecule 1 (STIM1), a calcium sensor localized in the endoplasmic reticulum [90–92]. Pyk-2-dependent phosphorylation of STIM1 is required for its interaction with calcium-release-activated calcium channel (Orai1) and Ca²⁺ entry from extracellular stores [92]. Importantly, phosphorylation of STIM1 at Y361 is required for the development of pulmonary edema in inflammatory lung [92]. Although studies in dermal microvascular endothelial cells isolated from neonatal foreskin challenge the role of Orai1 in agonist-induced barrier disruption [93], further understanding of Orai1 function with respect to intracellular signaling might explain tissue-specific role of Orai1 in microvascular endothelial cells.

The role of calcium-dependent phosphatases

The ion Ca²⁺ also enables the activation of calcineurin (CaN), a serine/threonine phosphatase [94, 95]. CaN dephosphorylates the microtubule end-binding protein 3 (EB3) at S162 allowing reorganization of the microtubule cytoskeleton [46]. It also dephosphorylates STIM1 and prolongs calcium entry from extracellular stores [96]. In addition, CaN rapidly dephosphorylates the transcription factor nuclear factor of activated T-cells (NFAT) protein to initiate NFAT nuclear import [94, 97, 98]. NFAT signaling in endothelial cells is linked to upregulation of transcripts associated with cytokine–cytokine receptor interaction [C-X-C motif chemokines (CXCL) 10 and 11] and the cell adhesion [vascular cell adhesion protein 1 (VCAM1)] pathways [52]. These factors facilitate entry of neutrophils into pulmonary circulation, augmenting vascular injury. Therefore, calcium signals also contribute to the inflammatory response by inducing expression of specific proteins in activated endothelium (Fig. 1).

Intracellular calcium-release channels in endothelium

Calcium homeostasis is finely regulated in endothelial cells through gating of selective and non-selective cation channels that enable cells to uptake, store, and release Ca²⁺ ions to manipulate their cytosolic concentration in a spatio-temporal manner [99]. The endoplasmic reticulum (ER) retains about 75% of total stored Ca²⁺ ions [100, 101], while other stores such as mitochondria, lysosomes, and endosomes hold about 25% [100, 102, 103], recently reviewed in [104]. Movement of Ca²⁺ ions in and out the cytosol is commonly regulated by channels and pumps [105–107], also reviewed in [108–111]. The channels, when open, enable rapid diffusion of the ion Ca²⁺ from intracellular or extracellular stores into the cytoplasm downhill concentration gradients. In contrast, the pumps move the ion Ca²⁺ in an energy-dependent manner against the concentration gradient (reviewed in [104, 112, 113]). The plasma membrane Ca²⁺ ATPase (PMCA) and the sodium calcium exchanger (NCX) both continuously remove the ion Ca²⁺ outside of the cells [114–116], whereas the sarco/endoplasmic reticulum Ca²⁺-ATPase (SERCA) moves the ion Ca²⁺ for storage into ER and the sarcoplasmic reticulum (SR) lumina [117]. These stores release the ion Ca²⁺ in response to extracellular stimuli to increase the ion Ca²⁺ concentration and induce calcium signaling in endothelial cells.

There two main types of Ca²⁺-selective intracellular channels, the inositol 1,4,5-trisphosphate receptors (IP₃Rs) and the ryanodine receptors (RyRs). IP₃Rs release Ca²⁺ ions from the ER, whereas RyRs release Ca²⁺ ions from the SR [118–126], as reviewed in [127, 128]. IP₃Rs are ubiquitously expressed IP₃-gated channels localized on the membrane of the ER [129]. These channels consist of four ~260 kDa subunits [130–133] that appear to be mostly mobile structures randomly organized into small clusters across the surface of the ER membrane [134–140]. Some evidence from patch clamp experiments using isolated nuclei suggests that IP₃R clusters of about four channels are spontaneously induced by IP₃ [141, 142]. Although the underlying mechanism of IP₃-evoked cluster formation remains unknown, the current view is that binding of IP₃ to the IP₃Rs may trigger conformational changes within the receptor that favor assembly of the clusters. Interestingly, at basal cytosolic Ca²⁺ concentrations, clustered IP₃Rs have a ~50% reduction in mean open time than lone IP₃Rs, suggesting that lone receptors are more active than clustered ones. However, the presence of free Ca²⁺ ions in cytosol reverses this inhibition [141]. The cytosolic portion of IP₃Rs contains a putative Ca²⁺-sensor region (residues 1933–2271), which may modulate properties of the channels. In the presence of Ca²⁺ ions, the clustered IP₃Rs demonstrate a higher open probability while showing reduced close time as compared to lone receptors; in addition, the receptors within the clusters exhibit simultaneous openings and closing [141], indicative of cooperative behavior. Therefore, the current model proposes that both secondary messengers, IP₃ and Ca²⁺ ion, bind to the IP₃Rs and attribute a positive regulation of Ca²⁺ release. Whereas IP₃ rapidly promotes assembly of IP₃R clusters, Ca²⁺ ions allow recruitment of Ca²⁺ release events through cooperative gating behavior of the receptors within the cluster.

IP₃ is generated by a class of membrane-associated enzymes, known as phospholipase C (PLCs), that hydrolyzes phosphatidylinositol 4,5-bisphosphate (PIP₂) [143, 144]. These enzymes are classified into six distinct families: PLC-β, γ, δ, ε, ζ, and η [145]. Eight out of thirteen known PLC isoforms have been identified in freshly isolated murine endothelial cells [146] from mesenteric arteries (PLCβ1, β3, and β4; PLCγ1 and γ2; PLCδ1 and δ3; and PLCε), whereas two additional isoforms have been detected in pulmonary (PLCη1 and η2) and middle cerebral arteries (PLCδ4 and PLCη1), indicating tissue-specific diversity of the phosphoinositide signaling pathways across the arterial beds. Whereas a function of PLCδ, PLCε, and PLCη isoforms is mainly unknown in endothelial cells, our previous work indicates that PLCγ1 is a major isoform responsible for maintenance of basal level of IP₃ in human lung microvascular endothelial cells [46]. This is consistent with the phenotype of PLCγ1 knock-out mice which die in utero due to loss of both erythroid and endothelial progenitor cells [147, 148]. In contrast, PLCγ2 is responsible for IP₃ generation downstream of VE–cadherin adhesion disruption, suggesting a potential role of PLCγ2 in endothelial cell proliferation and migration downstream of inflammatory and pro-angiogenic stimuli [46]. PLCβ isoforms act downstream of G-protein-coupled receptors (GPCRs) coupled to the G-proteins G_q and G₁₁ [149–152] or Vascular Endothelial Growth Factor (VEGF) Receptor 2 signaling [153].

Given the fact that sustained elevation of cytosolic Ca²⁺ is toxic to cells, several negative feedback mechanisms regulating the channel activity have evolved [154, 155]. As the intracellular ion Ca²⁺ concentration exceeds a threshold value of 0.92 mM, release of the Ca²⁺ ions is inhibited due to the Ca²⁺ binding to the inhibitory sites of IP₃Rs [156–159]. To date, seven Ca²⁺-binding sites have been discovered in the cytosolic portion of IP₃R1 [160–163], yet their significance in IP₃R gating remains unclear. Mutagenesis analysis of residues overlapping these Ca²⁺-binding sites has no significant effect on IP₃R gating [164]. Another model proposes that luminal concentration of Ca²⁺ is directly linked to sensitivity of IP₃Rs to the Ca²⁺ ions [165]. As the luminal concentration of Ca²⁺ falls, IP₃R loses its sensitivity to IP₃ even in the presence of cytosolic IP₃ and Ca²⁺. This model, however, needs to be supported with further experiments.

It has also been proposed that several proteins including Ca²⁺-binding proteins containing the tetra EF-hand, calmodulin (CaM), and the calmodulin (CaM)-like neuronal Ca²⁺-binding proteins (CaBPs), attribute to a negative regulation of the IP₃Rs [166–170], recently reviewed in [171]. These proteins bind IP₃R1 within the first 128 amino acid sequence [170, 172] and inhibit binding of IP₃ to the receptor at 0.15 mM free Ca²⁺ ions in cytosol leading to blocked Ca²⁺ release [166, 169]. Hence, cytosolic Ca²⁺ ions attribute to both positive and negative feedback regulations of Ca²⁺ release through IP₃Rs.

Another type of Ca²⁺-selective intracellular channel, the ryanodine receptors (RyRs) form Ca²⁺ channels located on the membrane of the SR. The tetrameric RyR channel consists of subunits ~5000 residues in size [173–176]. There exist three isoforms of RyRs in mammals with RyR1 and RyR2 being preferentially expressed in skeletal muscles and cardiomyocytes [174, 177, 178]. Endothelial cells predominantly express RyR3 isoform [179, 180]. RyRs activity is regulated by ryanodine, a plant alkaloid from Ryania speciosa [181, 182], which induces channel pore opening at nanomolar concentrations and promotes pore closing at micromolar concentrations [154, 183–185]. Furthermore, Ca²⁺ ions and ryanodine demonstrate cooperative behavior [186, 187]. An increase in free cytosolic Ca²⁺ facilitates stronger binding of ryanodine to the RyRs, as shown in endothelial cells [188]. In this respect, RyRs and IP₃Rs share analogous regulation of Ca²⁺ release by cytosolic Ca²⁺. Since our review is mainly focused on the function of IP₃Rs in endothelial cells, we refer to recent publications for additional information on structure and function of RyRs [189–191].

Recent studies have also revealed the presence of intracellular two-pore channels (TPCs) in endothelial cells that belong to a family of the voltage-gated ion channels [192–194]. The family is presented by three distantly related proteins (TPC1–3) with TPC2 specifically localized to the lysosomal and TPC1 to the endolysosomal systems [192, 195]. TPC3 is also targeted to acidic organelles as well as the plasma membrane [196, 197], but is lost in human and rodent species [198]. The crystal structure of a two-pore channel from Arabidopsis thaliana suggests that Ca²⁺ and membrane potential activate two distinct six transmembrane (6-TM) domains [199], indicating potential mechanism for Ca²⁺ and voltage sensing. It also proposes that luminal Ca²⁺ might stabilize the second voltage-sensing domain in the resting state and, thereby, shift voltage activation towards more positive potentials.

In endothelial cells, TPCs are mainly involved in regulation of cell proliferation [193, 194]. TPC2, which promotes Ca²⁺ release from lysosomal stores, potentiates pro-angiogenic effect of VEGF on endothelial cells, suggesting a specific role of TPC2 in VEGFR2-mediated angiogenesis [193]. In contrast, TPC1 is involved in intracellular calcium release from endolysosomal store in response to arachidonic acid and mediates specific calcium signaling associated with proliferation of circulating endothelial colony forming cells [194].

Function of IP₃R channels in endothelium

IP₃R isoforms

Through cDNA cloning on mammalian cells, three isoforms of IP₃Rs (IP₃R1, IP₃R2, and IP₃R3) have been discovered and characterized in various cell culture models [118, 135, 200, 201]. Even though all three isoforms demonstrate about 65–85% sequence similarity, they each produce a different magnitude of Ca²⁺ release due to their distinct affinities for IP₃ [131, 202–205]. The in vitro studies have shown that IP₃R2 is the most sensitive to IP₃ [135, 206] and produces long-lasting and regular Ca²⁺ oscillations [203, 207, 208]. IP₃R3, the least sensitive for this secondary messenger, instigates monophasic Ca²⁺ transients [203, 206, 209]. IP₃R1 demonstrates intermediate affinity for IP₃ and generates Ca²⁺ oscillations, which appear more irregular when compared with IP₃R2-mediated oscillations [206, 207, 209]. These differences in binding affinities for IP₃ among receptor types can be further modulated through the interaction of the N-terminus ligand-binding domain (LBD; see “Organization and structure of IP₃Rs”) with adenosine triphosphate (ATP), the regulatory proteins—such as CaM, CaMBP, CaN, and STIM 1—and homer protein homolog (HOMER) proteins, as well as through the receptor phosphorylation ([166–170, 209–218], recently reviewed in [128]).

Expression of endothelial IP₃R isoforms varies among the different vascular systems [53, 219–221]. Whereas all three isoforms are expressed in the endothelium of the aorta and arteries of systemic circulation in rats and domestic cattle [220, 222–226], IP₃R1 is not present in the pulmonary endothelium of either human or mouse [53]. Interestingly, some striking difference in the expression pattern of IP₃Rs is also noted among species. This difference might be attributed to specialized function of the receptor isoform in human since in IP₃R1 is predominant in endothelial cells of human brain microvasculature and mesenteric artery [219, 220], while IP₃R3 and IP₃R2 are preferentially expressed in pulmonary microvascular, and aorta endothelial cells [53, 221] as well as in circulating endothelial colony forming cells [227].

The role of IP₃R1 in regulating myogenic vascular tone

Interestingly enough, endothelial cell-specific knockout of IP₃R1 causes high blood pressure in mice due to disruption of NFAT/endothelial nitric oxide synthase (eNOS) signaling [228]. Consistent with this finding, other studies demonstrate that expression of IP₃R1 in aortic endothelial cell is downregulated in spontaneously hypertensive rats [222]. Both IP₃R1s and IP₃R2s are localized at myo-endothelial projections between endothelial and smooth muscle cells and selectively regulates inter-cellular communications. They are also involved in the control of vascular tone by activating intermediate conductance of calcium-activated potassium channels, thereby hyperpolarizing adjacent smooth muscle cells [224, 229, 230]. It is well known that fluid shear stress induces intracellular Ca²⁺ oscillations through activation of heterotrimeric G_q/G₁₁ proteins, which is coupled to purinergic receptor [231, 232] or bound to the plasma membrane in endothelial cells [233, 234]. Deficiency in IP₃R1 seems to disrupt these adaptive responses to increased blood flow in arteries of systemic circulation and to cause hypertension [228, 235, 236]. Cumulatively, these findings suggest that IP₃R1-dependent Ca²⁺ handling might play a critical role in regulating myogenic vascular tone in arteries. It is plausible that IP₃R1 is involved in physiological adaptive responses to hemodynamic changes in endothelium of systemic circulation, whereas it is dispensable in the low-pressure system such as pulmonary circulation.

The role of IP₃R2 and IP₃R3 in regulating endothelial permeability in lung

The other IP₃R isoforms, IP₃R2 and IP₃R3, expressed in endothelial cells have implications in both increased endothelial permeability and cytokine production during inflammation in the lung [52, 53]. Pro-inflammatory mediators such as serine protease α-thrombin and histamine induce Ca²⁺ ion release from ER stores. These mediators bind to and activate respective G-protein-coupled receptors (GPCRs), protease-activated receptor 1 (PAR-1) [237, 238], or histamine receptor 1 (H1) [239, 240], at the surface of endothelial cells (Fig. 2).

Fig. 2.

Model of receptor-mediated signaling in activating Ca²⁺ release and entry in endothelial cells during inflammation. The signal transduction begins with the binding of the ligand (hormone, cytokine, etc.) to specific G-protein coupled receptor (GPCR) at the surface of endothelial cells. Initial Gα_11/Gα_q-coupled activation of phospholipase C (PLC) β directs cleavage of phosphatidylinositol 4,5-bisphosphate (PIP₂) into the two secondary messengers, diacylglycerol (DAG) and inositol trisphosphate (IP₃). IP₃ induces clustering and activation of IP₃Rs on the ER membrane, whereas DAG instigates Ca²⁺ entry through transient receptor potential cation channels member 6 (TRPC6) located at the plasma membrane. The latter might contribute to increases in cytosolic Ca²⁺ concentration igniting global Ca²⁺ release through IP₃Rs. Depletion of the ER Ca²⁺ stores triggers oligomerization of stromal interaction molecule 1 (STIM1) which is required for activation of TRPC1 channel and Ca²⁺ entry from extracellular space. Lipopolysaccharide (LPS) also can activate Ca²⁺ entry through the TRPC6 channel by binding to toll-like receptor 4 (TLR4)

IP₃ binds to IP₃Rs and initiates Ca²⁺ release from ER stores [120, 241, 242]. Depletion of intracellular stores, in turn, activates a set of intracellular events that mediate capacitive or store-operated Ca²⁺ entry (CCE or SOCE) and, thus, replenish Ca²⁺ in ER (Fig. 2) [243, 244]. The level of Ca²⁺ ions within ER is sensed by transmembrane proteins localized on the ER membrane, STIM1 [90, 245, 246]. STIM1 binds to the Ca²⁺ ions through EF-hand motif at the N-terminus, a portion of the protein located in the ER lumen [247, 248]. Depletion of Ca²⁺ inside the ER leads to oligomerization as well as interaction of STIM1 with both the calcium-release-activated calcium channel proteins (Orai) and transient receptor potential channel 1 (TRPC1) located at the plasma membrane of endothelial cells [249–255].

As mentioned above, distinct stimuli can engage specific IP₃R isoforms in activated endothelium. IP₃R3 triggers monophasic Ca²⁺ transients in response to pro-inflammatory mediators such as α-thrombin, and promotes increased permeability of endothelial barrier [53]. IP₃R2 induces Ca²⁺ oscillations in response to endotoxin lipopolysaccharide (LPS), which is found in the outer membrane of Gram-negative bacteria [52]. This isoform is responsible for sustained activation of NFAT-mediated transcripts encoding the proteins associated with inflammation [52, 256]. In addition, DAG promotes Ca²⁺ influx through receptor-operated Ca²⁺ channels (ROC), the pathway unique from SOCE (Fig. 2) [257–260]. DAG interacts with TRPC6 at the plasma membrane [259] and permits Ca²⁺ entry through this non-selective cation channel in endothelial cells [49, 51, 261]. DAG can also contribute to inflammatory responses in endothelial cells through direct activation of protein kinase C (PKC) [44, 47, 59, 262–264] or generation of lipid secondary messengers such as leukotrienes [265–267]. Thus, calcium signaling emanated at the level of ER channels is amplified by Ca²⁺ flux from extracellular stores through SOCE and ROCE mechanisms.

Organization and structure of IP₃Rs

A potential mechanism of channel gating has been suggested based on the recently solved quaternary structure of the IP₃R1 channel in non-conducting state [133]. The IP₃R1 tetrameric structure consists of four subunits organized around a central axis [133]. Each subunit consists of two β-trefoil domains (β-TF1 and β-TF2), three armadillo solenoid folds (ARM1–ARM3), α-helical domain (HD), intervening lateral domain (ILD), six α-helices (TM1–TM6 or TMD), C-terminal domain (CTD), and a helical linker domain (LNK), which connects the transmembrane and the cytosolic bundles of each subunit (Fig. 3a, b).

Fig. 3.

Domain structure of IP₃R1. a Schematic representation of IP₃R1 domains with the corresponding amino acids. b Organization of individual IP₃R1 molecule within tetrameric channel; different domains are indicated by color as in a. c Structure of the central core of IP₃R1 channel (tetramer); axis is indicated by the dashed line. d Bundle of TM6 helices forming the conduction pathway of the IP₃R1 channel; four individual TM6 domains are highlighted by color. The P-helices (P) lining the luminal vestibule and selectivity filter (SF) loops are indicated. The pore is formed by chain of hydrophobic residues as indicated by color on the right. β-TF1 and β-TF2 β-trefoil domains, ARM1–ARM3 armadillo solenoid folds, HD α-helical domain, ILD intervening lateral domain, TM1–TM6 or TMD α-helices, CTD C-terminal domain, LNK a helical linker domain.

Modified from Fan et al. [133] with permission of the publisher. Copyright ©2015, Nature Publishing Group

Organization of the permeation pathway

The proposed model suggests that the TMD region is essential for Ca²⁺ ion conduction [133]. The Ca²⁺ permeation path is lined by four TM6 α-helices, which are oriented at 37^o with respect to the ER membrane (Fig. 3c). A luminal loop between TMD5 and TMD6 α-helices assembles a luminal vestibule. This loop contains both a short P-helix and a selectivity filter comprised of highly conserved residues [133, 268]. An important feature of a luminal vestibule is the presence of a positively charged ring, which is formed by the His2541 residues located within the P-helix of a tetrameric channel (Fig. 3c). This ring is predicted to repel the positively charged Ca²⁺ ions in the non-conducting channel. In conductive state, the P-helices are anticipated to undergo a structural rearrangement allowing passage of Ca²⁺ ions beyond a luminal vestibule [133]. The gateway for ion permeation is located at the constriction spot between four of the TM6 helices that each contains several hydrophobic residues oriented towards the permeation pathway (Fig. 3c). In the non-conducting channel, these hydrophobic residues form a pore of 5 Å in diameter that restricts passage of the hydrated Ca²⁺ ion of 8–10 Å in diameter [133]. In contrast to a luminal vestibule, a cytosolic vestibule formed above the pore is comprised of negatively charged residues, which are anticipated to promote a translocation of the Ca²⁺ ions into the cytosol [133].

Function of the cytosolic region

The cytosolic part of IP₃R1 contains two β-trefoil domains, three ARMs, and an α-helical domain (Fig. 3a, b). The modular organization of armadillo folds predicts their role in assembling different interfaces for binding of IP₃, Ca²⁺ ion, and the regulatory proteins [171]. The β-TF1 domain at the N-terminus (residues 5–225) constitutes a suppressor domain, whereas β-TF2 and two α-helices armadillo repeats of ARM1 assemble the IP₃-binding core (IBC) region. These three domains form a triangular structure, also known as ligand-binding domain (LBD) [126, 269, 270], above the CTD bundle. The α-helical, ARM2, and ARM3 domains connect the LBD to the channel-forming region. Even though the exact mechanism underlying IP₃-evoked conformational changes within the pore is not fully understood, it is proposed that gating of IP₃Rs occurs through a long-range communication between the LBD region and the pore [133].

The current model suggests that the binding of IP₃ to the IBC causes a closure of IP₃-binding pocket and, sequentially, a twist-like movement of the suppressor domain [133]. The model indicates that the interactions between the suppressor and ARM3 domains carry the signal from the LBD to the pore. Interestingly, the suppressor domain also establishes interaction with β-TF2 and ARM2 domains of adjacent subunit, suggesting that movement of the suppressor domain might transmit the signals to adjacent subunits. Consistent with this model, deletion of the suppressor domain or mutation of the Y167, site involved in the interaction with β-TF2, prevents IP₃-evoked gating of the channel [271, 272], indicating an importance of the suppressor domain in the transmission of long-range allosteric changes.

It has also been proposed that CTD domain plays an important role in gating of the IP₃R channel. The studies using IP₃R mutant lacking the 43 residues from the CTD domain have demonstrated that it is required for channel gating [273]. The CTD domain is comprised of the long helical bundle spanning the cytosolic portion of the receptor [133]. Both biochemical and structural studies indicate that the N-terminal and C-terminal parts of the receptor are closely associated with respect to each other [274]. The quaternary structure of the receptor reveals the presence of the electrostatic interactions between the CTD and β-TF2 domains from the adjacent subunits. The CTD also interacts with LNK, the bridge between the cytosolic and transmembrane parts of the receptor, and possibly transmits the signal to the TMD6 [133].

Therefore, ligand-evoked gating of the channel relies on both extensive inter- and intra-subunit interactions within the receptor comprised of four subunits. The first near-atomic resolution structure of the IP₃R1 [133] reveals conservation of the ion-conduction pore among tetrameric channels [275]. The specific features of the IP3R channels are carried through additional domains attached to the pore. A unique architecture in the CTD domain suggests a distinctive mechanism of IP₃R gating that requires direct coupling between the CTD and LBD domains of adjacent subunits [133].

IP₃-evoked Ca²⁺ release events

The fundamental question is “How does the ion Ca²⁺ elicit an array of signaling events alternating from proliferation, survival, and death?” Because diffusion of the ion Ca²⁺ is altered due to binding to numerous proteins and lipids in the cytosol, in many instances, it is only effective in initiating signal transduction events locally, in the proximity of the ion Ca²⁺ release or entry. The ion Ca²⁺ is released from intracellular stores through IP₃Rs in three major events called blips, puffs, and waves (Fig. 4). A blip is the elementary release event, a result of a passive diffusion of the Ca²⁺ ions from the ER through the lone channel (Fig. 4) [276]. This allows movement of the ion Ca²⁺ along the gradient as the result of a stochastic and autonomous process occurring at intermittent IP₃R sites across the ER [276, 277]. Blips typically last for 10 ms and are responsible for basal cytosolic Ca²⁺ concentration of about 40–80 nM in resting endothelial cells [53].

Fig. 4.

Elementary Ca²⁺ release events evoked by IP₃. Ca²⁺ release occurs in three major events: blip, puff, and wave. Blips represent release of Ca²⁺ through a single IP₃R channel, which can be lone or part of the cluster. Puffs are characterized by coordinated opening of multiple channels organized into the cluster. Regenerative Ca²⁺ waves are attributed to activation of IP₃R across the entire cell. Both IP₃ and Ca²⁺ ignite Ca²⁺ release form neighboring IP₃R clusters

A greater release of Ca²⁺ ions leads to puffs, a result of simultaneous openings of 5–6 channels grouped into the clusters lasting for about 100 ms [278–283]. As opposed to blip events, puffs originate within the clusters, suggesting that engagement of lone IP₃R channels into multichannel assembly plays a critical role in transition from blip to puff (Fig. 4) [284, 285]. The clusters igniting puffs are ~50 nm wide and distributed at ~3 μm from each other [286]. Multiple studies have experimentally established a direct correlation between frequency of puffs ignition and the number of IP₃Rs present within the cluster [276, 285, 287, 288]. This relationship suggests that larger IP₃R clusters have a disproportional influence on intracellular Ca²⁺ signaling and may act as an important ignition switch for regenerative Ca²⁺ waves (Fig. 4). However, it should be considered that blips and puffs do not solely serve as components of a global Ca²⁺ release; rather, they are signaling events in themselves.

As IP₃ concentration increases, puffs can trigger regenerative Ca²⁺ waves that spread across cell [286, 289] due to coordinated opening of a numerous channels of IP₃Rs within the ER (Fig. 4). It has been proposed that Ca²⁺ ions trigger higher frequency puffs through activation of neighboring channels. The current model suggests that puff frequency might set up a threshold for regenerative Ca²⁺ waves, where greater oscillation frequency attributes to higher excitability of the nearest IP₃Rs and thus the transition from local (puffs) to global (wave) calcium-release events [290–294]. In this respect, organization and distribution of IP₃R clusters become critical features. Below, we discuss the mechanisms regulating the receptor clustering in light of our recent work [53].

Regulation of IP₃R clustering in endothelial cells

Dynamics of IP₃R clusters in endothelial cells

Regenerative ion Ca²⁺ release relies on recruitment of neighboring IP₃Rs via a process known as Ca²⁺-induced Ca²⁺ release (CICR). Therefore, propagation of Ca²⁺ signals is critically dependent on organization of the IP₃Rs as well as optimal distribution of the receptors within the ER membranes. IP₃Rs assemble the clusters in unstimulated pulmonary endothelial cells [53] as well as other cell types [134, 135, 139]. IP₃R3s form less abundant clusters than IP₃R2s in endothelial monolayers, which can be explained by their different binding affinities to IP₃. This suggests that local spikes in basal IP₃ production that are mediated by PLCγ1 [46, 295] are sufficient to promote clustering of both IP₃Rs in unstimulated endothelial cells [53].

Consistent with IP₃R distribution in other cell types [134, 135, 138, 139], GFP-IP₃R3 form highly mobile structures in resting endothelial monolayers [53]. The IP₃R3 clusters appear to be very infrequent structures within the ER membrane that undergo spontaneous assembly and disassembly. Furthermore, consistent with hypothesis of “hot spots” for calcium release [277, 296, 297], some clusters demonstrate sporadic assembly and disassembly at the same location, perhaps in response to changes in local concentrations of IP₃. Furthermore, stimulation of the endothelial cells with α-thrombin promptly induces de novo formation of IP₃R3 clusters, which demonstrates exponential growth in size [53]. Our recent work identified the microtubule-associated protein EB3 as a key regulator of formation of IP₃R3 clusters in endothelial cells [53].

The role of microtubule cytoskeleton in IP₃R clustering

The role of the microtubule cytoskeleton in organization of IP₃-evoked Ca²⁺ signals has been established in various cell types [53, 225, 298–303]. IP₃Rs are shown to establish the transient associations with microtubules [53, 304], which in turn, actively participate in the redistribution of IP₃Rs within the ER membrane [300, 301, 305]. The microtubule cytoskeleton also contributes to initiation and propagation of Ca²⁺ waves in endothelial cells [225, 299] that may occur through spatial organization and temporal dynamics of IP₃R3s within ER membranes [53, 306, 307]. Furthermore, few reports indicate that the microtubule cytoskeleton facilitates delivery of IP₃ to IP₃Rs [308, 309], possibly contributing to initiation of Ca²⁺ spikes in secretory cells [310]. More recent studies in endothelial cells indicate that microtubules establish transient interactions with IP₃R3 localized at the ER membrane and play a critical role in assembling agonist-evoked IP₃R3 clusters [53].

While it is possible that the microtubule cytoskeleton promotes calcium signaling through multiple independent mechanisms, it has become apparent that it contributes to spatio-temporal organization of calcium release from ER stores. Microtubule dynamic is critical to this function because accumulation of specific proteins at the growing microtubule plus ends might provide the mechanism for tethering or stabilizing IP₃R clusters at ER membrane.

EB3 is a major regulator of IP₃R clusters in stimulated endothelial cells

The EB family consists of three members: EB1, EB2, and EB3 [311]. The EB proteins have demonstrated a high affinity for the outer part of growing microtubules, so-called the ‘+tips’ [312–314]. EBs are involved in establishing a protein network at the ‘+tips’ through recruitment of other proteins. Specifically, EB proteins interact with the CAP-Gly domain proteins (such as cytoplasmic linker and p150-Glued proteins) and proteins containing the short EB-binding motif S/TxIP [315–320].

IP₃Rs contain an EB-binding motif within a short unstructured region of the α-helical domain (HD), which connects the LBD to the channel-forming region [133]. The S/TxIP-containing motif conserved among mammalian IP₃Rs enables the interaction of IP₃Rs with EB3 and, albeit, a much weaker interaction with EB1 [53]. Intriguingly, EB3, but not EB1, is required for α-thrombin-evoked clustering and gating of IP₃R3 in endothelial cells. Depletion of EB3 or disruption of the IP₃R3 binding interface through a single-point mutation (T804 to A) within EB-binding motif of IP₃R3 reduces the number of the IP₃R3 (T804A) clusters in response to α-thrombin challenge [53]. The formed clusters are short-lived [53], suggesting that tethering of IP₃R3s to microtubule +tips by EB3 might play an essential role in stabilization of IP₃R3 intramolecular associations within the cluster. Disruption of this interaction alters organization of IP₃R3 clusters and thereby inhibits propagation of calcium signals in endothelial cells.

Our proposed model also suggests that formation of IP₃R3 clusters is essential for recruitment of neighboring IP₃Rs via Ca²⁺-induced Ca²⁺ release since IP₃-evoked Ca²⁺ signals are massively attenuated in EB3-deficient cells. The importance of EB3 in calcium signaling is clear from analyses of isolated lungs, where knockout of EB3 in endothelial cells protects lungs from increased vascular leakage and pulmonary edema [53]. Thus, the microtubule cytoskeleton provides a critical mechanism for assembling stable IP₃R clusters and effective Ca²⁺ signals in endothelium in the context of inflammation. Targeting this interaction with specific inhibitors might provide an attractive strategy for treatment of vascular leakage and pulmonary edema.

Conclusion remarks and future directions

Intracellular Ca²⁺ signaling is one of the major mechanisms involved in the communication between endothelial cells and the environment. Changes in mechanical (pressure and shear stress) and chemical (growth factors, hormones, and cytokines) stimuli are processed by endothelial cells through receptor-mediated signaling. The ion Ca²⁺ is a second messenger that is released from ER stores and ignites a myriad of signal transduction pathways upon the activation of endothelial cell surface receptors. The hierarchy of IP₃-evoked Ca²⁺ release events is predicted to differentially regulate the cellular responses. While elementary Ca²⁺ signals in the form of blips and puffs might be critical to regulation of vascular tone, globalization of Ca²⁺ signals through regenerative Ca²⁺ waves might contribute to vascular inflammation and leakage. Understanding a spatial–temporal organization of Ca²⁺ release in endothelial cells of intact blood vessels remains a major challenge in the field of endothelial biology.

Dynamic reorganization of IP₃Rs within ER membrane is crucial in igniting neighboring channels and transitioning from local to global calcium-release events. The microtubule cytoskeleton contributes to assembly of stable IP₃R cluster and sets up a positive-feedback loop for the amplification of Ca²⁺ signals associated with the progression of several pathologies including vascular leakage and inflammation. Increasing our understanding of IP₃R regulation at the molecular level will lay a ground work for the design of therapeutic interventions that target specific function of the receptor associated with pathology while leaving intact physiologically important function.

Compliance with ethical standards

Financial support for the author(s)

Supported by NIH Grants R01 HL103922 to Y.A.K; AHA AWARD 13PRE17090090 to M.G.

Conflict of interest

The authors declare that they have no competing interests.