I. INTRODUCTION

Top Previous Next References

Endothelial cells (EC) form a multifunctional signal-transducing surface that performs different tasks dependent on its localization in the vessel tree. Arterial EC provide a pathway for delivery of oxygen from blood to tissue. They modulate the tone of vascular smooth muscle cells (VSM), which in turn controls blood pressure and blood flow by adjusting the caliber of arteries and arterioles. In the microvascular bed, EC regulate the permeation of various metabolites, macromolecules and gases, as well as autocrine and paracrine factors and are involved in the regulation of cell nutrition. In all vessel types, EC are involved in blood coagulation, control of the transport between blood and tissue, movement of cells adhering to EC, wound healing, and angiogenesis. Several of these functions depend on constitutive properties of EC, e.g., presenting an anticoagulative surface for blood flow by secreting heparin sulfates and expressing ectonucleotidases and thrombomodulin (32, 33, 273, 295). Other functions require an active response of EC to various signals of mechanical, chemical, or neuronal nature. These signals originate from the blood or tissue side of EC or from a coupling with other cells, such as neighboring EC and adhering cells such as white blood cells and thrombocytes, but also with cells at the abluminal side, e.g., smooth muscle cells, mast cells, and fibroblasts. This signal transduction is impaired during vessel disease (arteriosclerosis) and injury, inflammation, or hemodynamic disturbances (hypertension).

The role of ion channels in the transduction of these signals into cellular responses is still a matter of debate and has received substantial attention only in the last few years. Ion channels are mainly involved in "short-term responses" (in the second and minute range) and affect the synthesis or release of pro- and anticoagulants, growth factors, and the release of vasomotor regulators. Some of these compounds are produced "on demand" [nitric oxide (NO), PGI₂, and platelet activating factor (PAF)], and others are released by exocytosis from storage granules [tissue plasminogen activator (tPA), tissue factor pathway inhibitor (TFPI), von Willebrand factor (vWF)] in response to various transmitters, such as acetylcholine, histamine, kinins (bradykinin), angiotensin, ATP, ADP, the coagulation factor thrombin, growth factors, but also in response to mechanical stimuli, such as shear stress and biaxial tensile stress. Fast responses related to the regulation of the permeability of the blood-tissue interface, controlled by changes in the contractile state of EC and their ability to modulate cell-cell contacts are also controlled by ion channels. The "slow responses" consist in the expression of several surface molecules, adhesion molecules, gene expression, cytoskeletal changes, remodeling, endothelial cell growth, and angiogenesis.

This review addresses the diversity of ion channels and their functional role in EC. Our current knowledge is limited to effects of ion channels on fast endothelial responses, these channels being mainly essential for the regulation of Ca²⁺ signaling. The major part of the review is therefore devoted to ion channels that function as pathways for Ca²⁺ entry. These channels are mainly responsible for a long-lasting increase in the free intracellular Ca²⁺ concentration ([Ca²⁺]_i) during various types of stimuli. This Ca²⁺ entry is given by

(1)

where J_Ca is the Ca²⁺ influx, F is the Faraday constant, and N is the number of channels. The open probability p of the channel and its conductance _Ca for Ca²⁺ represent inherent features of Ca²⁺ entry channels, which are discussed in the first part of this review. In addition, Ca²⁺ entry is regulated by the driving force for Ca²⁺, i.e., the difference between membrane potential (V_M) and equilibrium potential for Ca²⁺ (E_Ca). Thus Ca²⁺ entry will be tuned by channels that modulate electrogenesis in EC, which will be discussed in the second part of the review. Figure 1 illustrates the dual function of ion channels on Ca²⁺ entry in EC. Agonist stimulation induces in the absence of extracellular Ca²⁺ only a fast release of intracellular Ca²⁺, which is accompanied by a transient hyperpolarization. In the presence of extracellular Ca²⁺ however, it also induces a plateaulike phase of [Ca²⁺]_i due to Ca²⁺ entry which leads membrane hyperpolarization and an increased driving force for Ca²⁺ influx. Figure 1 gives also an overview of the most important channels in EC that are likely responsible for Ca²⁺ entry and modulation of membrane potential. The third part of this review discusses endothelial functions in which ion channels are likely involved.

View larger version (22K): [in this window] [in a new window]   Fig. 1. Dual function of ion channels for Ca2+ entry in endothelial cells. Simultaneous recordings of intracellular Ca2+ concentration ([Ca2+]i) and membrane potential (VM) in a freshly isolated mouse aorta endothelial cell in the presence and absence of extracellular Ca2+. Note the fast Ca2+ peak followed by a long-lasting plateau, which disappears after washing out the agonist (10 µM ACh, 37°C). This plateau is absent when extracellular Ca2+ is removed. Thus the functionally important plateau phase is mediated by Ca2+ entry. The increase in [Ca2+]i is accompanied by a hyperpolarization, which is maintained during the plateau phase, but transient in the absence of extracellular Ca2+ ([Ca2+]ext). This hyperpolarization stabilizes the driving force for Ca2+ entry, VM  ECa (see text). Candidates for the Ca2+ entry pathway are listed at the right together with channels that mainly control the driving force for Ca2+ entry (see text for further explanation). RACC, receptor-activated cation channel; CNG, cyclic nucleotide-gated channel; HCN, hyperpolarization and cyclic nucleotide-gated channel; CCE, capacitative Ca2+ entry; SOC, store-operated channel; NCX, Na+/Ca2+ exchanger; NSC, nonselective cation channel; ClCa, Ca2+-activated Cl channel; VRAC, volume-regulated anion channel; CFTR, cystic fibrosis transmembrane conductance regulator.

	II. VARIABILITY OF ION CHANNEL EXPRESSION IN ENDOTHELIAL CELLS

Top Previous Next References

EC express a bewildering variety of ion channels. Unfortunately, it is not clear which channels may be of functional impact in the vascular tree in vivo because some of these channels have only been detected in cultured endothelial cell lines or in primary cultured cells, others in cells obtained by diverse cell dispersion methods, and only very few from in situ measurements in blood vessels. It is therefore not surprising that the extensive amount of data on channel physiology in EC is sometimes controversial and often questionable with regard to their physiological significance. In the future, it might therefore be necessary to identify the candidate channel in vivo. An elegant method to identify channels in endothelium by cell-specific single cell RT-PCR has been published recently (337).

Another problem typical for EC is that ion channels in EC adapt their appearance to the external conditions. Apparently, ion channels or generally gene expression in EC are highly malleable and vary with cell isolation, culture, and growth conditions (136). As a striking example, human capillary EC grown in normal medium show TEA-sensitive outwardly rectifying K⁺ currents and inwardly rectifying K⁺ channels (IRK), but when grown in a "conditioned medium" (low cell density, medium changed only during passages at days 5-6 compared with controls in which the medium was changed every second day) expressed different channels, such as slo BK_Ca or 4-aminopyridine-sensitive A channels (173). The quantitative features of volume-regulated anion currents in primary explants of EC are different, i.e., activation during cell swelling is delayed and inactivation at positive potentials is accelerated as compared with cultured EC (381). The expression of IRK channels in freshly isolated human umbilical vein EC (HUVEC) changes with the number of passages (290). The loss of different serpentine receptors in EC during culturing is well documented (389). Likely, EC might be especially predisposed for this variability.

EC also show regional heterogeneity, including differences in [Ca²⁺]_i signaling, in immunological and metabolic properties, and in the pattern of released mediators for endothelium-dependent vasorelaxation (202, 245, 273, 386, 393).

The uncertainty as to which types of ion channels are present in a vessel type under physiological conditions is a complicating factor for a comprehension of ion channel function in EC.

	III. CALCIUM ENTRY CHANNELS

Top Previous Next References

Their role in Ca²⁺ signaling is obviously a fundamental aspect of ion channels in EC. We therefore first describe the various influx routes of Ca²⁺ in EC. Electrophysiological properties of the presently known endothelial Ca²⁺ entry pathways are summarized in Table 1.

Nonselective cation channels (NSC) poorly discriminating between cations have been detected in several ion channel families. There was so far no compelling evidence for the existence of NSC members of the purinergic ligand-gated receptor-channel complexes (the P_2X family) in EC, but very recently P2X4 receptors have been identified in several macrovascular primary cultured EC (448). This receptor, which is involved in ATP and shear stress induced Ca²⁺ influx (447), seems to be part of a far more complex Ca²⁺ entry system, because the ATP-induced Ca²⁺ entry still persists in HUVECs treated with antisense P2X4. NSCs belonging to the mammalian degenerins (MDEG) family are apparently not expressed in EC.

Various functionally different receptor-activated cation channels (RACCs), gated by binding of agonists to their membrane receptor, have been described in EC (123, 172, 177, 291, 295). Most of these channels are activated via signaling cascades involving activation of phospholipase C (PLC), but it is not clear which second messengers are involved. Interestingly, TRP3 and -6, which are members of the family of six transmembrane helix channels (STRPC, see below) that are present in endothelium (98), are nonselective RACCs (132). Probably, TRP4 and -5 are also nonselective RACCs (357), although it has been shown that TRP4 might be at least part of a more Ca²⁺-selective, store-operated channel (99). Endothelial RACCs, activated by an increase in [Ca²⁺]_i, are also permeable for Ca²⁺ and provide a Ca²⁺-entry route that exerts a positive feedback on their own activation. Their molecular structure is not known, although members of the TRP family with Ca²⁺-calmodulin binding sites in their primary structure (see sect. IIIC) are possible candidates.

Ca²⁺-permeable NSC activated by vasoactive agonists, which have been already described in freshly isolated cells from umbilical vein (270, 271, 276, 289, 291), have recently been characterized in more detail in cultured cell lines. The channel has a conductance of ~25 pS and is permeable for Na⁺, Cs⁺, and Ca²⁺ with a permeation ratio P_Ca:P_Na that varies between 0.03 and 2 (172, 177, 267, 272, 296). The current slowly activates during agonist stimulation and has been observed only in the presence of physiological [Ca²⁺]_i. Buffering of [Ca²⁺]_i with 10 mM BAPTA completely prevents current activation. On the other hand, loading the cells with Ca²⁺ via the patch pipette does not activate the current. Interestingly, application of a store-depleting inhibitor of sarco/endoplasmic reticulum Ca²⁺-ATPase (SERCA) pumps, such as tert-butyl-benzohydroquinone (tBHQ) or thapsigargin (TG), can also activate this NSC. The influx of Ca²⁺ through this NSC is coupled to agonist stimulation and depends on inositol 1,4,5-trisphosphate [Ins(1,4,5)P₃] production. Block of PLC with the pyrrole-dione derivative U73122 rapidly inhibits Ca²⁺ influx, whereas the pyrrolidine-dione derivative U73343, which does not inhibit PLC, is ineffective. Also 5-nitro-2-(3-phenylpropylamino)benzoic acid (NPPB), Ni²⁺, ecanozole, and SKF-96365 inhibit the agonist-induced Ca²⁺ entry (177). This channel shares some properties with the hTRP3 channels expressed in EC that do not express this NSC (178). Baron et al. (13) described another nonselective Ca²⁺-permeable cation channel, activated by intracellular Ca²⁺ and with a conductance of 44 pS for monovalent cations. This channel, with a permeability ratio P_Ca:P_Na of 0.7, is half-maximally activated at 0.7 µM [Ca²⁺]_i. A nonselective, agonist-induced current, which is also activated by [Ca²⁺]_i and suppressed by inhibitors of the cyclooxygenase pathway, has been described in aortic EC (137).

An amiloride-sensitive, poorly selective cation channel (23 pS for Na⁺ and K⁺) has been observed in brain microvessels (416). This channel seems to be important for regulation of cation fluxes across the blood-brain barrier. Constitutively open, nonselective Ca²⁺-permeable cation channels have been detected in both luminal and abluminal membranes of endocardial EC. They consist mainly of two types of nonselective outwardly rectifying cation channels with a single-channel conductance of 36 and 11 pS, respectively. The permeability sequence for monovalent cations is K⁺ > Rb⁺ > Cs⁺ > Na⁺ > Li⁺ which corresponds to the Eisenmann sequence IV and indicates a weak-field strength binding site for these cations (83). Gd³⁺ and La³⁺ are efficient blockers, whereas Ca²⁺ and Ba²⁺ but neither Mg²⁺ nor Mn²⁺ are permeable. Vasoactive agonists induce a nonselective cation current in endocardial EC with different ion selectivity and rectification properties (234, 235).

Oxidant stress, which induces superoxide anions, activates a 28-pS NSC in a calf pulmonary artery cell line. This channel is equally permeable for Na⁺ and K⁺ and also for Ca²⁺. Oxidized glutathione, a cytosolic product of oxidant metabolism, activates these channels, whereas reduced glutathione reverses activation (198). This channel opens in two gating modes that do not depend on intracellular Ca²⁺ stores or on [Ca²⁺]_i. Activation of these channels and the concomitant membrane depolarization may limit Ca²⁺ influx (199). TRP3 has been proposed as a molecular candidate for this channel (10).

A CNG1 member of the CNG family has been recently cloned in vascular endothelium. It is expressed in aorta, medium-sized mesenteric arteries, and small mesenteric arteries and forms a nonselective cation channel mediating entry of extracellular Ca²⁺ that might be involved in the regulation of arterial blood pressure (452).

Agonist stimulation of rat pulmonary artery EC in primary culture activates CNG channels, which resemble CNG2 subunits of rat olfactory neurons. This activation is subsequent to Ca²⁺ entry, probably via store-operated or RACC channels, and is prevented in the presence of specific inhibitors of soluble guanylate cyclase. It has been proposed that the increase of [Ca²⁺]_i upon agonist stimulation activates endothelial NO synthase (eNOS), which in turn enhances NO levels, stimulates soluble guanylate cyclase, and increases cGMP levels, which then leads to activation of CNG2 (see Fig. 2 and Ref. 441). This endothelial CNG is neither selective for K⁺, Na⁺, Cs⁺, nor Rb⁺. Its activation contributes to membrane depolarization and exerts a negative feedback on Ca²⁺ entry. Modulation of CNG by cAMP via G protein-coupled receptors (GPCR) has so far not been described in EC. It has to be mentioned that cGMP may also induce a negative feedback on SOC/RACC via a PKG pathway (see sect. IIIB).

View larger version (23K): [in this window] [in a new window]   Fig. 2. Negative feedback of CNG-induced membrane depolarization on Ca2+ entry. Ca2+ entry via RACC and/or SOC elevates [Ca2+]i and stimulates endothelial nitric oxide synthase (eNOS). The subsequent activation of soluble guanylate cyclase (sGC) increases cGMP. cGMP and cAMP [via agonist activation of G protein-coupled receptors (GPCRs), Gs, and adenylate cyclase (AC)] activate CNG and/or HCN channels, which induces membrane depolarization. This depolarization exerts a negative feedback on Ca2+ entry via RACC/SOC. In addition, a feedback inhibition of Ca2+ entry channels via activation of PKG has been described (203).

In summary, EC lack typical ligand-gated channels, such as the P_2X receptor. They express several Ca²⁺-permeable, receptor-operated NSC with diverse activation mechanisms, superoxide-gated NSC, and cyclic nucleotide-gated channels. Background NSCs are probably constitutively open. Some RACCs might be members of the STRPC family.

It is well known that compounds, such as tBHQ, cyclopiazonic acid (CPA), and TG, which deplete intracellular Ca²⁺ stores without affecting Ins(1,4,5)P₃ production, activate a Ca²⁺ influx pathway in EC (73, for a review see Refs. 295, 358, 359, 398, 399). This influx is apparently controlled by the filling degree of intracellular Ca²⁺ stores and is presumed to represent the major pathway for Ca²⁺ influx during agonist stimulation (318). Charged or filled stores prevent Ca²⁺ entry, whereas discharged or empty stores promote influx of extracellular Ca²⁺. We will use the term SOC for electrophysiological data and the term CCE for data derived from changes in intracellular Ca²⁺ upon readmission of extracellular Ca²⁺ to store-depleted cells. This distinction is not only semantic since the latter signals also depend on other mechanisms, such as Ca²⁺ sequestration and buffering which makes their interpretation often controversial. CCE signals are not much affected by mitochondrial Ca²⁺ buffering, but inhibition of the plasma membrane Ca²⁺ pump (PMCA) inhibits their secondary decay phase, indicating a modulating role of PMCA (194). Other reports also indicate a modulating role of a Na⁺/Ca²⁺ exchanger on CCE signals: inhibition of the exchanger increases CCE and delays the decay of store depletion-induced Ca²⁺ signals after reapplication of extracellular Ca²⁺ (75, 118, 315). Changes in membrane potential in non-voltage-clamped cells will also affect the driving force for Ca²⁺ and hence the CCE signals.

Store-operated Ca²⁺ channels (SOC) are obviously distinct from classical voltage-operated Ca²⁺ channels and Ca²⁺-permeable RACCs. In general, SOCs are much more selective for Ca²⁺ than the above-described Ca²⁺ entry channels. Only two highly Ca²⁺-selective agonist-activated channels have been described in EC. The one is activated by ATP and bradykinin and has been characterized at both the whole cell and single-channel level in bovine aorta EC (BAEC) (227). It is very selective for Ca²⁺ and Mn²⁺ and has a conductance of 2.5 pS for Mn²⁺. Inositol 1,3,4,5-tetrakisphosphate [Ins(1,3,4,5)P₄], but not Ins(1,4,5)P₃ increases its open probability, which is ~50% lower at 1 µM [Ca²⁺]_i than at 1 mM [Ca²⁺]_i. The other highly Ca²⁺-selective channel is gated by Ins(1,4,5)P₃ and has been observed in cell-attached, inside-out, and outside-out patches in BAEC. Ins(1,4,5)P₃ increases its open probability at a constant [Ca²⁺]_i, whereas Ins(3,4,5,6)P₄ is ineffective. This channel is reversibly blocked by heparin added to the intracellular side of inside-out patches and might be directly gated by Ins(1,4,5)P₃ binding (400).

The best-described SOC is the so-called "Ca²⁺-release-activated Ca²⁺-channel" (CRAC), which has been studied extensively in mast cells and T lymphocytes (145, 146, 318). CRAC is a highly Ca²⁺-selective channel with a single-channel conductance of ~30 fS for Ca²⁺ and of ~40 pS for monovalent cations under divalent-free conditions (188). The first "CRAC-like" current in endothelium was published already in 1992. Bradykinin activates a whole cell current in BAEC (~10 pA/cell at a potential of 60 mV) with characteristics similar to CRAC (251), which is blocked by La³⁺ and inactivates rapidly if Ca²⁺ is the main charge carrier but not if it is carried by Na⁺. Inwardly rectifying CRAC-like currents have also been recorded in other endothelial cells after depleting intracellular Ca²⁺ stores by dialyzing them with either a high concentration of BAPTA to buffer [Ca²⁺]_i at very low concentrations or with Ins(1,4,5)P₃, by extracellular application of ionomycin, or administration of the SERCA blockers TG or tBHQ. (88, 109, 110, 306). Store depletion protocols also activate Ca²⁺ channels in BAEC with a conductance between 5 and 11 pS and which are not highly selective for Ca²⁺, i.e., P_Ca:P_Na = 10 (394, 398, 399). The density of this current is approximately 0.5 pA/pF at 0 mV, which is ~10-20 times smaller than that reported for Jurkat and peritoneal mast cells (318). Its permeability sequence is Ca²⁺ > Na⁺ > Cs⁺ and Ca²⁺ > Ba²⁺ (88). Elevation of extracellular Ca²⁺ increases the current amplitude, whereas removal of extracellular divalent cations induces a large Na⁺ current that is blocked by micromolar concentrations of La³⁺.

In general, it seems that SOC currents in EC exhibit much weaker inward rectification, are less Ca²⁺ selective, and have a higher single-channel conductance than the typical CRAC currents. A typical example for activation of a SOC current in EA.hy926 cells, a cell line derived from human umbilical vein, is shown in Figure 3. Breaking into the cell with a patch pipette containing a high concentration of BAPTA, 30 µM Ins(1,4,5)P₃, and 20 mM tBHQ activates within 30-60 s an inward current, which reverses at a potential more positive than 50 mV and is completely blocked by 1 µM La³⁺. This protocol bypasses the agonist-mediated activation pathway via PLC activation described for RACC, but the exact activation pathway of this channel remains to be elucidated.

View larger version (22K): [in this window] [in a new window]   Fig. 3. Store depletion activates a cation current in mouse aorta endothelial cells (MAEC). A: activation of Ca2+ entry can occur via depletion of intracellular Ca2+ stores. The mechanism of activation is unknown and may involve a diffusible messenger, direct coupling between SOC and the inositol 1,4,5-trisphosphate receptor (IP3R), or store-dependent exocytosis and membrane incorporation. B: breaking into an isolated MAEC under the conditions shown in the inset of C activates a cation current. The current is blocked by 1 µM La3+ and shows a typical delayed unblock. C: the current reverses around +60 mV, indicating a Ca2+-selective current. (From S. H. Suh and B. Nilius, unpublished data.)

Pharmacological data on SOC currents in EC are rather scarce. They are almost completely blocked by 1 µM La³⁺ and inhibited by econazol, NPPB, Ni²⁺, and carboxy-amidotriazole (CAI), but a selective pharmacological blocker is not available. It has recently been shown that metabolites of the sphingomyelin pathway inhibit CRAC in RBL cells (244). However, sphingosylphosphorylcholine, which increases membrane capacitance in RBL cells, activates SOCs and NO production in EC (257). Likely, membrane lipids and metabolites of membrane breakdown are directly involved in regulation of CCE (see also sect. IIIC). The thiazolidinedione troglitazone and the imidazole-derivative SKF 96365 both block CCE in EC (186, 210, 366). It has been shown recently that binding of capsaicin to vanilloid receptor (VR) receptors, which are members of the OTRPC subfamily, inhibits CCE (57).

Mechanisms of CCE modulation are also still not much elaborated. Stimulation of EC with vasoactive compounds, and application of TG results in tyrosine phosphorylation of 42- and 44-kDa isoforms of mitogen-activated protein kinase (MAP kinase). Inhibition of this phosphorylation attenuates CCE. Tyrosine kinases may therefore be associated with the regulation of CCE (94). Calmodulin seems to inhibit CCE (394), whereas activation of protein kinase C (PKC) inhibits CRAC (318). Activation of protein kinase G (PKG) has been shown (204) to inhibit both SOC (patch-clamp data) and CCE (Ca²⁺ entry).

The signal transduction mechanism linking changes in intraluminal Ca²⁺ to the opening of plasma membrane Ca²⁺ channels is still controversial. Various second messengers have been suggested, such as a low-molecular-weight "Ca²⁺ influx factor" (CIF, 500 Da), an unidentified tyrosine kinase or phosphatase, the arachidonic acid metabolite 5,6-epoxyeicosatrienoic acid (5,6-EET), cGMP, 1,2-diacylglycerol (DAG) or metabolized arachidonic acid derivatives to explain the cross talk between Ca²⁺ stores and membrane channels. Activation of cytochrome P-450 epoxygenase 2C, the endothelial isoform of the endoplasmic reticulum cytochrome P-450 monooxygenase (CytP-450 MO), by -naphthoflavone induces the biosynthesis of 5,6-EET and potentiates CCE in BAEC, porcine aorta and human umbilical vein (119), whereas several P-450 inhibitors diminish CCE and induction of CytP-450 MO. In addition, depletion of endothelial Ins(1,4,5)P₃-sensitive stores directly activates CytP-450 MO (139). It has therefore been proposed that CytP-450 MO-derived 5,6-EETs may constitute a signal for CCE activation.

It has also been suggested that Ins(1,3,4,5)P₄ or Ins(1,4,5)P₃ or a decrease of [Ca²⁺]_i in a restricted subplasmalemmal space may directly gate the entry channels and that a direct coupling between CRAC and the Ins(1,4,5)P₃ receptor may control channel gating (8, 14, 58, 141, 144, 193, 318, 342). A direct physical contact between Ins(1,4,5)P₃ receptors in the endoplasmic reticulum and COOH-terminal contact sites of the putative plasma membrane TRP channels may be functionally involved (192, 193). Two short stretches of 54 (position 742-795) and 21 amino acids (position 777-797) that bind strongly to specific sequences in the Ins(1,4,5)P₃ receptor have been identified in hTRP3. The binding domain in the InsP₃ receptor comprises 289 amino acids downstream of the Ins(1,4,5)P₃ binding site (position 638-926) (37). This protein-protein complex may fulfill a similar function as the dTRP, PKC, receptor complex with the PDZ-domain protein InaD in Drosophila (258, 339). More recently, it has been shown that stimulation of vascular endothelial growth factor (VEGF) receptor-2 by VEGF-E, which induces store depletion probably via phosphorylation of PLC-, activates CRAC-like currents , which are blocked by micromolar concentrations of La³⁺ and are enhanced in amplitude if carried by Na⁺ in the absence of extracellular Ca²⁺ (single-channel conductance of ~45 pS; Lepple-Wienhues, personal communication), a value similar to that observed in T lymphocytes (36-40 pS, Ref. 188).

The tremendous efforts in the last 3 years to identify mammalian genes encoding for Ca²⁺ entry channels have resulted in the discovery of the trp gene family ("transient receptor potential") in photoreceptors of Drosophila. The genes of this still expanding family might be related to several types of Ca²⁺-entry channels but in a much broader sense than encoding for store-operated Ca²⁺ entry channels only (22, 29, 144, 457-459). Trp genes encode a superfamily of proteins with six transmembrane helices, which can be divided into three subfamilies, i.e., S(hort)TRPCs, L(ong)TRPCs, and O(sm-9-like)TRPCs (Fig. 4, Ref. 132). The latter group, named after its homology to the osm-9 members in Caenorhabditis elegans which respond to changes in osmolarity, contains interesting members, such as the epithelial Ca²⁺ channel (ECaC) involved in Ca²⁺ reabsorption in the distal tubules of the kidney, NSC such as the vanilloid receptor VR1, VRL1, and the insulin-like growth factor regulated channel GRC (132, 140, 183, 408). The LTRPC group contains proteins of more than 1,600 amino acids, the function of which is presently not understood.

View larger version (26K): [in this window] [in a new window]   Fig. 4. The superfamily of TRP channels. Putative domain structure of the TRP proteins consisting of 6 transmembrane helices. The NH2 terminus contains several ankyrin repeats that may be important for interaction with the cytoskeleton; the COOH terminus contains a proline-rich motif. The putative pore region is between helix 5 and 6. This basic structure is common to all members of the TRP channel superfamily (131). This family can be divided on the basis of homology into three subfamilies: short (STRPC), osm (OTRPC), and long (LTRPC). These subfamilies have structural differences, as indicated. For further explanation, see text and References 131, 141.

The STRPC group of proteins between 700 and 1,000 amino acids comprises the mammalian isoforms of TRP1-7, which are composed of six transmembrane helices with a pore-forming loop between the fifth and sixth helix (Fig. 4). The NH₂ terminus of all isoforms contains three highly conserved ankyrin domains. The NH₂-terminal part of the COOH terminus bears a proline-rich motif (29, 132, 329, 333, 457). The TM4 region lacks the positive charges typical for voltage-gated ion channels (16, 29, 457). STRPCs contain several calmodulin-binding regions, which may mediate the Ca²⁺-dependent modulation of store-related Ca²⁺ entry. However, the functional effect of calmodulin is difficult to evaluate because it influences many EC functions including the uptake of Ca²⁺ into intracellular stores (437). Some STRPCs (e.g., TRP1) have NH₂-terminal coiled-coil structures and COOH-terminal dystrophin motifs (for details see Ref. 333). STRPCs are mainly activated by agonist stimulation of diverse receptors coupled to different PLC isoforms but some probably by store depletion.

It is not clear how these channel proteins are organized, but the extreme functional heterogeneity observed in expression experiments may point to a multiheteromeric structure. Proper heteromerization may depend on unidentified auxiliary proteins similar to the Drosophila InaD (from the mutant "inactivation no afterpotential") protein. This protein contains five PDZ domains (comprising DTSL motifs) and is physically linked to the rhodopsin PLC (NorpA), a PKC (InaC), and the G_q-protein photoactivated by rhodopsin. The entire structure forms a signaling complex in which dTRP and dTRPL are integrated (258). PDZ domains probably mediate the clustering of membrane and membrane-associated proteins with TRP channels. A similar protein hINAl (human InaD like), which is twice as long as the Drosophila InaD (1,525 vs. 675 amino acids), has been cloned in human (330) and has been detected in human EC (V. Flockerzi and B. Nilius, unpublished data). In this context, formation of structural complexes of Ca²⁺ entry channels could also be integrin dependent, since an integrin-associated protein (IAP-50) physically bound to integrin ₃ has been proposed to function as a Ca²⁺ entry channel (363). This signaling pathway seems to be linked to the generation of inositol phosphate and the production of Ins(1,4,5)P₃ (328). INAF is another 241-amino acid protein that is probably involved in Ca²⁺ signaling via STRPC proteins, since mutations in the gene encoding for this protein directly affect TRP channel function (215).

The activation steps following PLC stimulation are unclear for all STRPC isoforms. One group of STRPCs (still frequently described as TRPs) seems to be activated via production of DAG and by binding of polyunsaturated fatty acids (PUFAs; Refs. 58, 132, 141, 142). Stimulation of EC via activation of the pertussis toxin-insensitive G_q protein-coupled PLC- causes Ins(1,4,5)P₃ production that initiates store depletion and probably STRPC-mediated Ca²⁺ entry. Also PLC-, which is activated by agonist-stimulated receptor tyrosine kinases and growth factor receptor tyrosine kinases, may activate STRPCs (98). Figure 5 summarizes the activation pathways for STRPCs that may be relevant for EC. It has to be noted that only a few members of the putative STRPC family have been analyzed at the functional level.

View larger version (28K): [in this window] [in a new window]   Fig. 5. Activation mechanisms of the short TRP channels. Short TRP channels are mainly activated via agonist receptors (G protein-coupled receptor, GPCR, and receptor tyrosine kinases, RTK) (131). These receptors activate various phospholipase C. Some STRPCs are directly activated by 1,2-diacylglycerol (DAG) or via still unknown interactions with G proteins, Ca2+, or even InsP3. TRP3, -6, and -7 are probably receptor-activated cation channels or RACCs. Another mechanism of activation is probably related to emptying of intracellular Ca2+ stores (see Fig. 3A).

Phylogenetic analysis of the mammalian STRPCs (TRPs) reveals four subfamilies (see Fig. 6A; Refs. 98, 132, 142, 333).

View larger version (26K): [in this window] [in a new window]   Fig. 6. Overview of the TRPC superfamily and expression pattern of STRPCs in mouse aorta endothelial cells (MAEC). A: phylogenetic tree of the mammalian STRPCs and the two members of Drosophila (TRP and TRP-like, TRPL). Functionally, four different types are discussed (for the controversial discussion see text). B: expression pattern in MAEC revealed by RT-PCR. The + and  refer to presence or absence of reversed transcriptase. GAPDH, glyceraldehyde 3-phosphate dehydrogenase. [From Freichel et al. (98). Copyright 1999 Karger, Basel.]

1) The first subfamily is a subclass that comprises TRP4 and TRP5. Heterogeneously expressed mTRP4 and mTRP5 are NSCs with a single-channel conductance of 42-66 pS that can be stimulated by guanosine 5'-O-(3-thiotriphosphate) (GTPS). The activation pathway may integrate signaling pathways from G protein-coupled receptors and receptor tyrosine kinases independently of store depletion. Their biophysical properties are inconsistent with those of CRAC channels. In other experiments, however, TRP4 and -5 are clearly store dependent and have a much higher Ca²⁺ selectivity (331, 332, 434).

2) TRP1, which is closely related to TRP4 and -5 in terms of evolutionary distance, probably forms SOCs (223, 224, 431) but is probably less Ca²⁺ selective than TRP4 and -5 (329, 387, 434).

3) The short TRP3, TRP6, and TRP7 with ~800 amino acids form store-independent NSCs probably activated via the DAG pathway (141) and are also sensitive to [Ca²⁺]_i (178, 307, 461). TRP3 and -6 have a conductance between 25 and 66 pS and are not very Ca²⁺ selective (141, 178). However, also a store-dependent activation mechanism has been considered for hTRP3 (124, 193).

4) The subfamily consisting of TRP2 is still uncertain and forms probably truncated channel proteins.

Several TRPs are expressed in endothelium, as illustrated by their expression pattern in mouse aorta in Figure 6B. Their expression varies, however, considerably between different types of EC, as shown in Table 2. BPAEC express TRP1 and -4; BAEC two splice variants of TRP1, -3, -4, and -5; HUVEC cells TRP1, -3, and -4; mouse aorta EC TRP1, -2, -3, -4, and -6; and rat pulmonary EC TRP1 but neither TRP3 nor -6 (52, 124, 178, 259).

SOC currents have been identified in BPAEC cells (88), but agonist stimulation does not activate Ca²⁺-permeable NSCs in these TRP3-deficient cells. Expression of hTRP3 gives rise to functional expression of 25-pS NSCs and enhances the agonist-induced Ca²⁺ increase in these cells (Fig. 7, Ref. 178), suggesting that hTRP3 channels may serve as a Ca²⁺-influx pathway during agonist stimulation. These channels share several properties with an endogenous nonselective Ca²⁺ entry channel observed in TRP3-expressing EA.hy926 cells (177). Expression of an NH₂-terminal fragment of hTRP3, which exerts a dominant negative effect on TRP channel function presumably due to suppression of channel assembly (124), abolishes activation of a store-dependent, agonist-activated nonselective cation current in HUVEC cells (124). These results suggest a role of STRPC-related proteins as store-operated and/or agonist-activated cation channels identical to TRP3. Other studies show that -estradiol significantly downregulates TRP4, and transretinoic acid dramatically upregulates TRP5, whereas these hormones have little effect on CCE (52). Expression of TRP1 and TRP4 in BPAEC induces a cation current that is activated by tBHQ and Ins(1,4,5)P₃, suggesting that both STRPCs may form functional but not highly Ca²⁺-selective channels (409). Interestingly, activation of TRP1 in rat pulmonary EC induces changes in EC shape and remodeling of the peripheral (cortical) filamentous actin (F-actin) cytoskeleton (259).

View larger version (18K): [in this window] [in a new window]   Fig. 7. Functional expression of hTRP3 in bovine pulmonary artery endothelial cells (BPAEC), which do not express TRP3. A and D: comparison of the effects of ATP (1 µM) on membrane potential of wild-type and htrp3 transfected BPAEC. Note the small hyperpolarization in the wild-type cells due to activation of a Ca2+-activated Cl channel. In hTRP3 expressing cells, a clear depolarization toward the reversal potential of a nonselective cation channel can be observed. Membrane potential was measured in current-clamp mode. B and E: changes in [Ca2+]i in nontransfected and htrp3-transfected cells. Note the much stronger increase in [Ca2+]i during agonist stimulation in the htrp3 transfected cell. C and F: current (I)-voltage (V) curves obtained from voltage ramps (100 to +100 mV) in both cells types before (a) and during application of ATP (b). Note the appearance of an IRK channel at negative potentials and a Cl channel at positive potentials (see also Fig. 11). In transfected cells, a clear activation of a cation current can be observed which correlates with the increased Ca2+ plateau. [From Kamouchi et al. (178), with permission from J Physiol.]

Another question refers to the nature of Ca²⁺ stores involved in CCE activation. In addition to Ins(1,4,5)P₃-mediated Ca²⁺ release, ryanodine-sensitive Ca²⁺ release (RsCR) seems also to be important for Ca²⁺ signaling in EC. Several studies have presented evidence for endothelial ryanodine receptors (1, 62, 213, 460). In human aortic EC, ryanodine-sensitive Ca²⁺ stores seem to be involved in the regulation of agonist-sensitive, Ins(1,4,5)P₃-depleted Ca²⁺ pools. Caffeine and ryanodine induce only a small Ca²⁺ entry if the Ins(1,4,5)P₃-sensitive stores are filled. However, after store depletion, caffeine but not ryanodine induces Ca²⁺ entry (62). One likely interpretation of these effects is that caffeine regulates CCE and/or that caffeine also controls the extent of store depletion by agonist stimulation. Also other experiments show that RsCR may modulate but not initiate CCE. By using deconvolution microscopy to monitor simultaneously cytosolic, perinuclear Ca²⁺ by fura 2 and membrane near, subplasmalemmal Ca²⁺ by the cell membrane impermeable dye FFP-18 in the same EC (315), it has been shown that Ca²⁺ release initiated by ryanodine increases subplasmalemmal Ca²⁺ but not perinuclear Ca²⁺. In addition, conditions that initiate RsCR also increase CCE. Although these data support the existence of local Ca²⁺ gradients, they are not indicative for CCE activation by the discharge of ryanodine-sensitive Ca²⁺ stores. Other findings do not support any effect of the discharge of ryanodine- sensitive stores on Ca²⁺ entry or Mn²⁺ quenching (354). However, subplasmalemmal Ca²⁺ elevation could dramatically change the driving force for Ca²⁺ entry by activating BK_Ca channels, which in turn would affect CCE (see sect. IVA2).

The modulation of endothelial Ca²⁺ signaling via a Na⁺/Ca²⁺ exchanger (NCX) is an incompletely resolved issue. Several groups have shown that a reduction of the Na⁺ gradient increases [Ca²⁺]_i via an NCX-mediated Ca²⁺ entry (352, 369, 378, 385). Evidence that the exchanger may also shape Ca²⁺ transients activated by vasoactive agonists comes from experiments in which [Ca²⁺]_i transients are modulated by changing Na⁺ gradients, loading EC with Na⁺ by using the ionophore monensin, application of NCX blockers, such as 3',4'-dichlorobenzamil (in comparison with the NCX ineffective amiloride) or La³⁺ (369). Recently, a NCX protein was detected in caveolin-rich EC membrane fractions (385).

	IV. ION CHANNELS CONTROLLING MEMBRANE POTENTIAL

Top Previous Next References

A crucial role of ion channels in endothelial Ca²⁺ signaling is the fine-tuning of the electrochemical gradient for Ca²⁺ (176, 226, 271, 272). The influence of membrane potential, which as discussed in section VIIIA is determined by the expression pattern of various ion channels, on Ca²⁺ entry in EC has been clearly demonstrated (170, 176, 226, 359). K⁺ channels, including Ca²⁺-activated K⁺ channels, inwardly rectifying K⁺ channels, and probably also voltage-dependent K⁺ channels, are the major class of ion channels in setting the membrane potential. In addition, it has been shown recently that Ca²⁺ entry in freshly isolated EC from rabbit aorta requires the presence of extracellular Cl to maintain a polarized membrane (433). Thus Cl channels, e.g., volume-regulated anion channels (VRAC), Ca²⁺-activated Cl channels (CaCC) and, interestingly, also cystic fibrosis transmembrane conductance regulator (CFTR) channels, recently detected in EC (388), might be important modulators of the inwardly driving force for Ca²⁺. In addition, NSC may tune the membrane potential of resting and activated EC (178, 419).

Some NSC are obviously impermeable for Ca²⁺ (419). They may induce a negative feedback on Ca²⁺ entry by inducing depolarization of EC (see also Fig. 2). Little information is available at the molecular level. Probably, some TRP members might be candidates for background Ca²⁺-impermeable NSC in resting endothelium.

Recently, a family of NSC has been cloned which includes channels activated by hyperpolarization and by cyclic nucleotides (HCN; Refs. 24, 25-27, 228). Four new members of this family with a length between 780 and >900 amino acids have been cloned recently. These channels consist of six transmembrane domains (TM) and are modulated by binding of cyclic nucleotides to a CNBD. TM4 is positively charged and contains at least 10 arginine and lysine residues. These channels show a high overall similarity with the cone CNG. They activate slowly at hyperpolarized potentials and binding of cyclic nucleotides. They are permeable for monovalent but not for divalent cations, Na⁺ being more permeable than K⁺ (P_Na:P_K ~0.15-0.3). HCN channels play a role as mediators of NO- and cGMP-dependent cellular processes (25). Therefore, they might be of potential importance in EC. Indeed, in EC from the blood-brain barrier, a hyperpolarization-activated current carried by Na⁺ and K⁺ has been described (165) that is modulated by NO, cGMP, and cAMP. Its modulation by a cyclic guanosine nucleotide may also explain previous results about a role for NO in regulating blood-brain barrier function.

B.  K⁺ Channels

1.  Inwardly rectifying K⁺ channels (IRK or K_ir)

IRK is one of the most important channels for the control of the resting potential in nonstimulated cells. It conducts inward currents at potentials more negative than the K⁺ equilibrium potential but permits much smaller currents at potentials positive to that potential. However, this channel is quite heterogeneously expressed in different EC types. Some macrovascular EC express the inward rectifier channel (BPAEC, BAEC, coronary EC), others do not (microvascular EC, EA.hy926 cells), whereas in other cells only a fraction of the cells shows IRK activity (HUVEC, Ref. 290). In general, the channel appears to be more abundant in cultured than in primary EC. In intact endocardial endothelium, it is mainly confined to the luminal side of the cells (235), but it appears to be randomly distributed in monolayers of cultured BAEC (60). If present, this conductance together with the basal activity of the volume-regulated anion channel and a background NSC determine the resting membrane potential of EC (46, 96, 419). As discussed in section VIIIA, the impact of these channels on the resting potential is often counteracted by that of Cl and background NSC channels.

In addition to its well-understood role in controlling electrogenesis, IRK is a K⁺ sensor, as shown in Figure 8. A small increase in extracellular K⁺, e.g., resulting from the activation of BK_Ca channels, shifts the reversal potential for K⁺ toward a more positive value and thereby increases the conductance of IRK. This enhanced IRK conductance is sufficient to overcome the depolarizing action of the nonselective cation and Cl conductance (see also Fig. 11) and shifts the membrane potential toward more negative potentials.

View larger version (26K): [in this window] [in a new window]   Fig. 8. IRK (Kir2.1) channels are K+ sensors. A: topological scheme of the EC Kir 2.1 channel, showing the GYG motif in the pore and the RRESEI phosphorylation site in the COOH terminus that is important for channel modulation. B: typical I-V curves in BPAEC (see also Figs. 7 and 11). The net current I-V relationships are composed of several ionic current as discussed above. Note the flat I-V relationship at low extracellular K+ concentrations, which are consistent with an unstable membrane potential. C: increasing extracellular K+ to 12 mM, although shifting the reversal potential of IRK (Kir2.1) toward more positive values, stabilizes the membrane potential at more negative values because of the increase in conductance of IRK. After washing out of the elevated K+, the bistable behavior of the membrane potential is reflected in the transient fluctuations between two polarization levels. D: effect of changing extracellular K+ from 6 to 12 mM in nine individual EC is also shown. Note the homogeneous response indicating a K+ sensor function of IRK. (From B. Nilius and J. Prenen, unpublished data.)

The conductance of single endothelial K_ir channels ranges from 23 to 30 pS in symmetrical K⁺ solutions (84, 156, 180, 290, 295, 325, 372, 373). Typical features of this channel are the increase in single-channel conductance with the square root of the extracellular K⁺ concentration (289, 373, 462) and a permeation profile of P_K > P_Rb > P_Cs (325, 373).

Extracellular Ba²⁺, TEA, TBA, and Cs⁺ all block K_ir (214, 322, 343, 419, 427). Extracellular Mg²⁺ causes a time-dependent block at negative potentials, which can be antagonized by extracellular K⁺. At positive potentials, also intracellular Mg²⁺ induces a time- and voltage-dependent block (84).

The shear stress evoked hyperpolarization of EC has been attributed to an enhanced outward whole cell K⁺ current, which can be blocked by Cs⁺. Shear stress also enhances the open probability of IRK channels in luminal cell-attached patches (312) and in inside-out patches (163).

IRK seems to be metabolically regulated. Intracellular ATP but not its nonhydrolyzable analogs adenosine 5'-O-(3-thiotriphosphate) (ATPS) and adenylyl imidodiphosphate (AMP-PNP) prevent its rundown in whole cell mode. Reduction of intracellular ATP and induction of hypoxic conditions dramatically downregulate IRK (180). The phosphatase inhibitor okadaic acid also prevents rundown, but protamine, an activator of phosphatase 2A (PP2A), enhances the rate of rundown. Phosphorylation of the channel molecule therefore seems to be essential for maintaining its activity, its rundown probably being due to dephosphorylation by PP2A (180).

An interesting observation is the inhibition of IRK in capillary and macrovascular endothelial cells by the vasoactive agonists angiotensin II, vasopressin, vasoactive intestinal polypeptide (VIP), endothelin (ET)-1, and histamine (150, 290, 322, 456). A similar inhibition has been observed if GTPS is applied via the patch pipette (150, 180). This inhibition is probably due to activation of a pertussis toxin (PTX)-insensitive G protein, which may mediate these inhibitory actions by modulating PPA2.

IRK has been identified as Kir2.1, a member of the Kir family (95, 179). Two highly conserved transmembrane regions, consisting of 427 amino acids and a highly conserved TIGYG-H5 motif in the pore region, characterize this channel. pH insensitivity, which is also characteristic for the endothelial IRK, is conferred on the channel by a methionine at site 84 (M84). Mg²⁺ and spermine, spermidine, and putrescine block appears to be linked to D172. S425 in the COOH-terminal phosphorylation motif RRESEI, which is important for modulation of Kir2.1, might be a candidate for the observed modulation by phosphatases (180).

In addition to well-defined IRK channels, a number of poorly characterized K⁺ channels, apparently not belonging to any of the cloned subfamilies of K⁺ channels, have been observed in EC. An example is the 170-pS inward rectifier in aorta EC activated by isoprenaline, adenosine, forskolin, and membrane-permeable analogs of cAMP and inhibited by PKA inhibitors (117).

The initial response to most EC stimuli is an increase in [Ca²⁺]_i that is often due to influx of Ca²⁺ and therefore depends on the electrochemical Ca²⁺ gradient, primarily set by the membrane potential. The depolarizing action of Ca²⁺ influx on the membrane potential is in various EC compensated by activation of Ca²⁺-dependent K⁺-channels (K_Ca), their increased open probability at elevated [Ca²⁺]_i causing membrane hyperpolarization. Different types of Ca²⁺-activated K⁺ channels have been described, e.g., BK_Ca or maxi-K channels, IK_Ca or intermediate-conductance channels, and SK_Ca or small-conductance channels. However, the expression of these channels between different types of EC is also very heterogeneous (295).

High-conductance Ca²⁺-activated K⁺ channels, BK_Ca, are expressed in most freshly isolated EC and EC in primary culture. Their conductance ranges between 165 and 240 pS (129, 147, 176, 179, 220, 289, 349, 381). The fast elevation in [Ca²⁺]_i during stimulation with vasoactive agonists induces a fast increase in the probability of the channel being open. The open probability is enhanced at more positive potentials, suggesting that the apparent Ca²⁺ affinity of the channel is voltage dependent and that the open probability depends on both [Ca²⁺]_i and voltage. An increase in [Ca²⁺]_i shifts the voltage dependence of activation toward more negative potentials (13, 63, 129, 176, 317).

The endothelial BK_Ca is blocked by charybdotoxin (IC₅₀ ~50 nM), iberiotoxin, TEA (IC₅₀ ~1 mM, 100% block at 10 mM), D-tubocurarine, and quinine. Extracellular Mg²⁺ block is voltage dependent (13, 349, 350). The benzimidazolone compounds NS004 and NS1619 (in the µM range) increase the open probability of BK_Ca in EC and shift its activation curve toward more negative potentials (121, 316). Nanomolar amounts of intracellular dehydrosoyasaponin (DHS-I) induce discrete episodes of high channel open probability interrupted by periods of apparently normal activity. DHS-I decreases the Ca²⁺ concentration for half-maximal activation of the channel (113, 247). BK_Ca activators may be functionally important, since it has been shown in many assays that inhibition of BK_Ca interferes with NO release (42). It has not yet been shown that cGMP-dependent kinases can activate BK_Ca in EC like in smooth muscle (166, 445).

Ca²⁺-activated K⁺ channels in vascular smooth muscle cells are targets for various physiological factors released from the endothelium, including NO (31, 346, 384) and endothelium-derived hyperpolarizing factor (EDHF; probably a cytochrome P-450-derived arachidonic acid metabolite; Refs. 92, 334). This latter compound also exerts an autocrine effect on BK_Ca in inside-out patches of primary cultures of endothelial cells from pig coronary artery (12). A similar autocrine effect of NO release is however unlikely, since the NO donor S-nitrosocysteine does neither directly activate nor modulate BK_Ca channels activated by increased [Ca²⁺]_i in cultured endothelial cells (129). Interestingly, K⁺ efflux via endothelial BK_Ca might activate K⁺-sensitive smooth muscle IRK channels and induce hyperpolarization. Thus K⁺ itself may act as EDHF (82).

BK_Ca has been identified by RT-PCR analysis as hslo in EA.hy926 cells (179). Mammalian slo channels, consisting of ~1,200 amino acids, share a high degree of homology with voltage-dependent K⁺ channels and consist of six transmembrane segments with a positively charged S4 segment. Five additional hydrophobic segments form a unique secondary structure in the COOH terminus of the mammalian slo channel. The origin of the voltage dependence is not completely clear, but hslo channels probably contain a Ca²⁺-independent intrinsic voltage sensor (411). An extra segment (segment 0) in the NH₂ terminal is probably a coupling site for the -subunit. Four -subunits have been cloned so far: ₁ seems to be mainly expressed in smooth muscle cells; ₂ in endocrine cells; ₃ probably in epithelium, liver, pancreas, and intestinal tract; and ₄ in the central nervous system (15). The unique COOH terminus possibly contains four additional helices, H7-H10, which represent almost 70% of the protein (80, 392). A 28-amino acid stretch between segment 9 and 10 contains highly conserved negatively charged amino acid residues that possibly form a highly selective Ca²⁺ binding site, the "Ca²⁺ bowl" (Fig. 9A, and Ref. 362). Expression of -hslo in EC that lack BK_Ca channels enhances the agonist-induced increase in [Ca²⁺]_i and shifts the membrane potential toward more negative values, a functional example in support of the role of the driving force for Ca²⁺ influx in EC. Figure 9, B and C, shows an example of such an expression experiment in BPAEC, which does not express functional slo channels. The plateau phase of the Ca²⁺ transient in BPAEC cells transfected with hslo is much more pronounced than in nontransfected cells. This finding correlates with the hyperpolarizing action of vasoactive agonists in the transfected cells and indicates that activation of hslo-BK_Ca exerts a positive feedback on the Ca²⁺ signals by counteracting the negative (depolarizing) effect of Ca²⁺-activated Cl channels (179).

View larger version (26K): [in this window] [in a new window]   Fig. 9. Functional expression of the -subunit of hslo in BPAEC, which does not express BKCa. A: topology of the hslo structure. The -subunit is probably not expressed in EC. B: changes in membrane potential and [Ca2+]i in a control BPAEC during stimulation with 1 µM ATP (Ca2+-activated Cl channels inhibited with 100 µM niflumic acid) are shown. [From Kamouchi et al. (178). Copyright Harcourt Brace & Co. Ltd. 1997. Reprinted with permission of Churchill Livingstone.] C: same as B, but in a BPAEC transfected with the -subunit of hslo. D: note the difference in resting potential, the more pronounced hyperpolarization, and enhanced plateau Ca2+ level during agonist stimulation in the transfected cell. [Modified from Kamouchi et al. (178).]

The -subunit of hslo could not be detected with RT-PCR methods in EC, which does not exclude the presence of other -subunits. The BK_Ca opener DHS-I, which only works in the presence of the -subunit, is ineffective in EC (248, 383), a functional indication for the absence of the ₁-subunit. Expression of the -subunit in EA.hy926 cells induces a leftward shift of the activation curve and has a sensitizing effect on BK_Ca (316, 317).

Ryanodine-sensitive Ca²⁺ release during agonist stimulation is the main activator of BK_Ca in an umbilical vein-derived cell line, suggesting that these channels may form a subplasmalemmal complex with Ca²⁺ release units located in their close vicinity (100).

Intermediate-conductance, Ca²⁺-activated K⁺-channels, IK_Ca, are inwardly rectifying and have a conductance between 30 and 80 pS in symmetrical K⁺ and 15 pS at physiological extracellular K⁺. IK_Ca are activated by Ins(1,4,5)P₃-sensitive Ca²⁺ release induced by agonists, such as bradykinin, acetylcholine, and ATP (355). ET-1 and ET-3 activate IK_Ca channels in brain microvascular endothelium via endothelin-A receptors (404). A Ca²⁺-dependent 30-pS channel in cultured BAEC is regulated by G proteins (401). GTP together with Mg²⁺, as well as GTPS stimulates IK_Ca, an effect that is reversed by guanosine 5'-O-(2-thiodiphosphate) (GDPS). An agonist- activated IK_Ca with a single-channel conductance of 34 pS and a dissociation constant (K_D) of 0. 63 µM for Ca²⁺ has been described in intact endocardial cells (234).

The pharmacology of the endothelial IK_Ca is not very detailed. Charybdotoxin, quinine, and TBA are efficient blockers of IK_Ca channels (220, 355, 356, 404). Noxiustoxin, a toxin purified from Centuroides scorpion toxin (61, 396), and interestingly NS1619 (44), an activator of BK_Ca channels, are potent inhibitors of IK_Ca channels. The hydrophilic oxidative reagents 5,5'-dithio-bis(2-nitrobenzoic acid) and thimerosal reduce IK_Ca channel activity in BAEC recorded in inside-out experiments but do not affect its unitary conductance. This activity is partly restored by the SH-group reducing agents dithiothreitol or reduced glutathione. The lipid-soluble oxidative agent 4,4'-dithiodipyridine is less potent, suggesting that critical SH groups localized at the inner face of the cell membrane may be involved in channel gating (45).

The biophysical and pharmacological profiles of these channels in EC are consistent with those of the recently cloned hIK (158, 168) and mIK (403) channel, which has an amino acid sequence related to, but distinct from, the six-TM helix, small-conductance SK_Ca channel subfamily (~50% conserved). Currents are inwardly rectifying, K_D for Ca²⁺ activation is 0.3 µM, and their single-channel conductance is 39 pS. The channels are reversibly blocked by charybdotoxin (inhibitory constant = 2.5 nM) and clotrimazole, minimally affected by apamin, iberiotoxin, or ketoconazole (158). Importantly, this channel lacks putative Ca²⁺-binding sites, but the proximal COOH terminus contains a Ca²⁺-dependent calmodulin (CaM)-binding site that is probably associated with four CaM molecules (87, 190).

Small-conductance Ca²⁺-activated K⁺ channels, SK_Ca, with a conductance of ~10 pS in asymmetrical conditions have also been observed in EC (122, 265, 353). Unlike BK_Ca, these channels are not voltage sensitive. They are blocked by TBA, apamin, clotrimazole, and D-tubocurarine (122). Recently, two types of SK_Ca channels were identified in excised patches of rat aortic EC with a conductance of 18 and 9 pS in symmetrical 150 mM K⁺, and 6.7 pS and 2.8 pS at physiological extracellular K⁺ concentrations (241). The smaller one is completely blocked by 10 nM apamin and 100 µM D-tubocurarine, and the larger one is insensitive to apamin but inhibited by charybdotoxin at concentrations >50 nM. The topology of SK_Ca is similar to that of voltage-gated K⁺ channels, exhibiting six TM helices and an intracellular COOH and NH₂ terminus. Primary sequences, however, are very different. Likely, four subunits form the channel. No EF hand, C2A, or Ca²⁺ bowl motif indicative for Ca²⁺ binding has been found. Gating of SK_Ca involves a Ca²⁺-independent interaction of calmodulin with the COOH-terminal domain of the -subunit and is mediated by binding of Ca²⁺ to the first and second EF hand motifs in the NH₂-terminal domain of CaM (187, 442).

The reduced intracellular ATP content during ischemic/hypoxic conditions can be mimicked by intracellular dialysis with ATP-free solutions or application of extracellular glucose-free/NaCN solutions. Under these conditions, increased whole cell and single-channel currents have been observed in EC from rat aorta and brain microvessels. Currents are reversibly blocked by glibenclamide or ATP in inside-out patches and activated by the K_ATP channel opener pinacidil (164). Lowering intracellular ATP or applying the K⁺ channel activator levcromakalim evokes unitary currents with a conductance of 25 pS in rabbit aortic EC. They are inhibited by glibenclamide or in inside-out patches by increasing the ATP concentration (184, 185). Low concentrations of the K⁺ channel openers diazoxide and rilmakalim cause a pronounced glibenclamide-sensitive hyperpolarization in coronary capillary EC and induce a rapid Ca²⁺ transient followed by a sustained elevation of [Ca²⁺]_i. The high sensitivity to diazoxide has been interpreted as a novel type of K_ATP channel composed of a novel combination of K_ir and sulfonylurea receptor (SUR) subunits (208). At present, Kir6.1 and Kir6.2 were detected in freshly isolated capillaries from guinea pig heart. It is not known whether Kir6.1 and/or Kir6.2 -subunits form the K_ATP channels together with the -subunit SUR2B in EC (249).

Langheinrich et al. (207) have also reported a glibenclamide-sensitive hyperpolarization induced by K⁺ channel openers and glucose deprivation in capillaries isolated from guinea pig hearts. Interestingly, evidence for a role