KATP channels are integral to the functions of many cells and tissues. The use of electrophysiological methods has allowed for a detailed characterization of KATP channels in terms of their biophysical properties, nucleotide sensitivities, and modification by pharmacological compounds. However, even though they were first described almost 25 years ago (Noma 1983, Trube and Hescheler 1984), the physiological and pathophysiological roles of these channels, and their regulation by complex biological systems, are only now emerging for many tissues. Even in tissues where their roles have been best defined, there are still many unanswered questions. This review aims to summarize the properties, molecular composition, and pharmacology of KATP channels in various cardiovascular components (atria, specialized conduction system, ventricles, smooth muscle, endothelium, and mitochondria). We will summarize the lessons learned from available genetic mouse models and address the known roles of KATP channels in cardiovascular pathologies and how genetic variation in KATP channel genes contribute to human disease.
I. THE ROAD TO THE DISCOVERY OF KATP CHANNELS
Since the first use of the voltage-clamp technique in cardiac tissue in 1964 (172), the description of the ionic nature of cardiac excitation was riddled with problems and artifacts, in part due to technical limitations of the recording equipment, such as the two-microelectrode and sucrose gap voltage-clamp techniques, and partly due to the complexity and multicellular nature of cardiac tissue (57). Two events coincided in the early 1980s that greatly accelerated the pace of cardiac electrophysiological discovery. The first was the successful isolation of enzymatically isolated, Ca2+-tolerant cardiac myocytes (96, 187, 643), and the second was the Nobel prize winning description of the tight-seal, single-electrode patch-clamp technique (316, 742). Subsequent progress was rapid, and many new discoveries were made, for example, a description of Ca2+ channels and their regulation by cAMP (102) and a description of a acetylcholine-activated K+ channel in nodal tissue (683). Other new advances in the early 1980s included the finding in the laboratory of Edward Carmeliet that metabolic inhibition by dinitrophenol (DNP) led to action potential shortening of isolated ventricular myocytes, due to activation of a time-independent outward K+ current (379, 832). This finding verified prior observations by the same group using conventional approaches made in multicellular preparations (radioactive flux studies and two-electrode voltage clamping), demonstrating that hypoxia induces K+ fluxes and an outward K+ current that are responsible for the action potential shortening (845, 848-850). The descriptions of the sarcolemmal/plasmalemmal ATP-sensitive K+ channel in ventricular myocytes by Noma (607) and Trube and Hescheler (815) gave unique insights into the unitary channel events that underlie this metabolically active outward K+ current. KATP channels have subsequently been found in most other tissue types, and we know that they have diverse roles in human physiology and pathophysiology (22). KATP channels have subsequently also been found in the inner membrane of mitochondria (376). For simplicity, throughout this review, we will refer to the class of sarcolemmal/plasmalemmal channels as “KATP channels,” and those in the mitochondria will be referred to as mitochondrial KATP channels.
Diverse classes of KATP channels exist, which differ significantly from each other in different tissues and cell types in terms of their nucleotide sensitivities, their biophysical properties, and their sensitivities to pharmacological agents (some examples are shown in Tables 1–3). The various subtypes of KATP channels originate in part from the assembly from specific combinations of the various Kir6.x and SURx subunits. For example, sarcolemmal KATP channels in striated muscle and those in pancreatic β-cells have the same pore-forming subunit (Kir6.2), but differ in the use of the regulatory subunit (SUR2A in the case of the ventricular channel and SUR1 in the pancreas). Smooth muscle KATP (or KNDP) channels, in contrast, appear to be composed of Kir6.1 subunits in combination with SUR2B subunits (721). The mitochondrial KATP channels are also sensitive to ATP and antidiabetic sulfonylureas. They are important in the setting of ischemia/reperfusion (291), but their molecular composition is only now being defined. The focus of this treatise is to review the properties of cardiovascular KATP channels and attempt to provide an impartial summary of the literature and views related to the roles for these channels in cardioprotection.
II. PROPERTIES OF CARDIOVASCULAR KATP CHANNELS
The following text is mainly focused on the KATP channel of the ventricular muscle, which has properties that are better characterized than the KATP channels in most other cardiovascular tissue types. Where possible, the ventricular KATP channels are compared with KATP channels in other cardiovascular tissues such as the atria, the specialized cardiac conduction system, the endothelium, the coronary vascular smooth muscle, and the mitochondria. For a more detailed description of these channels, please refer to section VII. The regulation of KATP channels at the protein and molecular level is dealt with in section IV. For description of the regulation of KATP channels by receptor signaling pathways, please refer to a recent review (806).
A. Intracellular ATP Blocks the KATP Channel
A defining characteristic of KATP channels is that their open probability is vastly decreased in the presence of ATP at the cytosolic face of an excised membrane patch (416, 607, 815) (Figure 1). Other types of K+ channels are mostly insensitive to cytosolic ATP, but certain types of K+ channels are activated by cytosolic ATP, such as the members of the renal Kir1 subfamily (e.g., the ROMK1 channel) and members of the retinal Kir4 subfamily (353, 784). For KATP channels, an inverse relationship exists between the channel's opening probability and the ATP concentration, which can be described by a modified Hill equation (432). In the initial studies performed with inside-out membrane patches obtained from cardiac ventricular myocytes, the ATP concentration needed to produce half-maximal block was ∼100 μM (607). The channel can also be blocked by nonhydrolyzable ATP analogs or in the absence of Mg2+, demonstrating that ATP hydrolysis is not required for channel block (23, 135, 416, 486). For example, AMP-PNP, AMP-PCP, and ATPγS (909) effectively block KATP channels from cardiac ventricular myocytes and pancreatic β-cells (23, 416, 613). Other nucleotides also inhibit KATP channel activity, but less efficiently. For example, the block by free ADP is ∼10–20 times less potent than that of ATP (23, 486, 754). Other nucleotides (e.g., CTP, GTP, ITP, or AMP) are even less effective (486, 754). In the cardiovascular system, the ATP sensitivity is higher in the atrium and ventricle than some other tissues (IC50 value: ventricle ∼ atrium < smooth muscle < conduction system ≪ mitochondria; Table 2).
B. MgADP Stimulates KATP Channel Activity: Regulation by the ATP/ADP Ratio
Although free ADP can block the KATP channel (albeit less effectively than ATP; see above), it became apparent that Mg2+ modulates the channel′s response to ADP. In the “open” cell-attached configuration, in which the cell is permeabilized by a brief exposure to saponin (in an attempt not to disrupt the cytosolic milieu, as would be the case with whole-cell recordings), MgADP was found to stimulate KATP channel opening (194) (Figure 2). This stimulation was not observed with AMP, cAMP, or adenosine. The nonhydrolyzable ADP analog ADPβS did not stimulate KATP channel activity, suggesting the involvement of phosphorylation processes. This experiment gave rise to the concept that the physiological control of KATP channel opening is mediated by changes in the ATP/ADP ratio, rather than solely by ATP (194). Subsequent experiments demonstrated that the stimulatory effects of ADP is Mg2+-dependent (227, 486), which gave further credence to the notion that phosphorylation and/or hydrolysis processes are involved.
C. AMP Stimulates KATP Channel Activity and Surface Expression
AMP, rather than ADP, is increasingly being recognized as the key regulatory molecule for sensing changes in energy metabolism and for regulating metabolic pathways (325). Intracellular AMP levels are almost undetectably low in normal well-oxygenated heart tissue due to the action of adenylate kinase, which interconverts ATP, ADP, and AMP and maintains their concentrations close to equilibrium according to this reaction: In well-oxygenated cells, the AMP levels are kept at very low levels since this reaction operates from right to left, ensuring an ADP/ATP ratio of ∼1:10 and an AMP/ATP ratio, which varies as the square of the ADP:ATP ratio, of ∼1:100 (327). When ATP consumption exceeds production during cellular stress, the ADP/ATP ratio rises, but the AMP/ATP ratio rises much more. For example, for a 5-fold increase of the ADP/ATP ratio, the AMP/ATP ratio will rise 25-fold (327). Thus the intracellular AMP concentration changes much more dramatically than that of ATP or ADP. Evolution has adapted to monitor changes in cellular energy status by responding to changes in AMP levels, for example, by activating AMP-activated protein kinase (AMPK), which has been described as the “fuel gauge of the mammalian cell” (326). Although AMP does not directly regulate KATP channel opening (607), there is now evidence that AMP-dependent activation of AMPK increases ventricular KATP channel open probability (905) and, similar to the situation in pancreatic β-cells (109, 632), enhances KATP channel surface density through trafficking mechanisms (767, 824). AMP may also participate in phosphotransfer reactions that are mediated by adenylate kinase to regulate the ADP/ATP ratio in the immediate vicinity of the channel to stimulate KATP channel opening (100). Thus, even though the action is indirect, elevated AMP levels during cellular stress function to stimulate KATP channel function (Figure 2).
D. Phosphorylation Is Required for Maintained KATP Channel Activity
In one of the earliest recordings of KATP channels from inside-out patches of ventricular myocytes, it was noted that many KATP channels becomes inactive within 5 min after patch excision (815). This phenomenon, referred to as “rundown” of KATP channels, was found paradoxically to be partially prevented (or even reversed) by MgATP, but not by AMP-PNP, ATPγS, or ATP in the absence of Mg2+ (226, 231, 613). This result suggested that phosphorylation processes are involved in maintaining channel activity following patch excision (822). Interestingly, the presence of the divalent cation Ca2+ on the cytosolic face of the membrane powerfully inactivates the KATP channel (or induces rundown) (226), and Ca2+-dependent inactivation can be restored by MgATP. Since the MgATP-dependent recovery is inhibited by wortmannin (884), it is believed that loss of specific phospholipids in the membrane after patch excision is partly responsible for the rundown process. It is not clear to what extent the rundown process is an experimental artifact or whether it has physiological implications. Nevertheless, these studies have contributed to our understanding of the role of specific inositolphoshate products in the regulation of KATP channel function (see sect. IVF).
E. KATP Channels Exhibit Weak Inward Rectification Properties
Many K+ channels are regulated by voltage and/or Ca2+ and are referred to as Kv or KCa channels (126). A group of K+ channels that is not regulated in this manner is referred to as inward rectifier K+ (Kir) channels, with the name due to the rectification properties and the shape of their current-voltage relationships. The Kir channels can be arbitrarily divided into a class with strong inward rectification properties or weak rectifiers, which exhibit larger outward currents upon membrane depolarization. KATP channels are members of the latter group (596). We will discuss the mechanisms of inward rectification in section IV.
III. THE PROTEIN SUBUNIT COMPOSITION OF KATP CHANNELS
Historically, the word channel refers to an ion conduction pathway (60), and its use in electrophysiology may have originated with the German word Kanäle, which means a “watercourse” or a “pipe.” By this definition, a “channel” (e.g., KATP channel) is an electrophysiological entity. A subunit is a biochemical entity and is a component of the protein complex that gives rise to channel function (e.g., Kir6.2 subunit) and a gene is the DNA that codes for the subunit (e.g., KCNJ11 gives rise to the Kir6.2 subunit) (Table 1). It has, however, become colloquial to use these terms interchangeably by referring to a channel as the ionic current, the protein, the mRNA, or even the gene. We will attempt to follow the traditional convention, but also use the short-hand notation (e.g., Kir6.2 channel).
B. Pore-Forming Subunits
Topologically, the Kv and KCa channels consist of protein subunits that span the membrane with six (or seven in the case of the BK channel′s subunit, KCa3.1) α-helixes (126). The Kir family of subunits, in contrast, consists of proteins with two membrane-spanning α-helixes (126, 597) (Figure 3A). Protein subunits of channels in the Kir family are grouped in several subfamilies (Kir1 to Kir7), with one or more paralogs within each subfamily (e.g., Kir3.1 to Kir3.4). Members of two of these subfamilies (Kir1 and Kir6) have been implicated as being the pore-forming subunits of KATP channels (Figures 4 AND 5).
1. Kir1.1 subunits
Not long after the identification of molecular components of voltage-gated K+ channels using molecular cloning methods (418, 631, 765, 785, 798), there was much excitement when the first subunit of an inward rectifier K+ channel (Kir1.1 or ROMK1; from the KCNJ1 gene) was identified by expression cloning using poly(A)+ RNA isolated from the outer medulla of rat kidneys (353). ROMK1 channel opening requires the presence of MgATP, but not Na2ATP, demonstrating the involvement of phosphorylation processes (554). ROMK1 is generally functionally associated with K+ fluxes in the kidney (860). In contrast to MgATP, Na2ATP at concentrations less than 10 mM has little effect on the activity of either the Kir1.1 channel expressed in oocytes (555) or the native kidney KATP channel (859). MgATP binds to the Kir1.1 COOH terminus with high affinity, e.g., TNP-ATP binding occurs with a Kd of ≈3 μM (836). Although activated by MgATP, there is evidence that MgATP at high concentrations can block Kir1.x channels, with ROMK2 being more sensitive to ATP block (with an IC50 of ∼5 mM) than ROMK1 (555). Consistent for a major role of Kir1.1 channels in the kidney, KCNJ1 mutations in humans are associated with Bartter syndrome (340), which is characterized by salt wasting, hypokalemic alkalosis, hypercalciuria, and a low blood pressure. Several mRNA splice variants have been described, giving rise to multiple protein variants (ROMK1 to ROMK6) (58, 72, 455). ROMK2 (or Kir1.1b) is a splice variant of Kcnj1, which is a subunit with a truncated NH2 terminus. It has been identified by expression cloning using poly(A)+ RNA from whole rat kidney and has a single-channel conductance of 30–36 pS (110, 930). As is the case for many inward rectifier K+ channels (126), the channel comprised of Kir1.1b is highly selective for K+ and is blocked by Rb+ and Ba2+. It is also blocked by oxidizing agents (244) and is weakly blocked by glibenclamide (930). The activity of ROMK1 channels is inhibited ∼10–15% by 200 μM glibenclamide (456), but the glibenclamide sensitivity is increased (IC50 = 0.6 μM) when ROMK2 is coexpressed in Xenopus oocytes with a Cl− channel, the cystic fibrosis transmembrane regulator (CFTR) (553). Although suggested previously (184, 790), SUR2B appears not to interact with Kir1.1b and is not required to confer the intrinsic low glibenclamide sensitivity to the channel (456). A recent study suggested that ROMK2 may contribute to the molecular composition of mitochondrial KATP channels. Specifically, ROMK2 contains an NH2-terminal mitochondrial targeting signal. Moreover, when heterologously expressed in H9C2 cells (a rat embryonic heart-derived cell line), a ROMK2 construct with a COOH-terminal tag was found to localize to intracellular compartments positive for mitochondrial markers (240). The same study also demonstrated that diazoxide-induced changes in mitochondrial matrix volume and thallium flux into mitochondria are abrogated by tertiapin Q, a peptide isolated from bee toxin that blocks certain types of inwardly rectifying K+ channels, including Kir1.x (ROMKx), G protein-activated K+ channels (Kir3.x or GIRK1/4) (404, 441), and the Ca2+-activated large-conductance K+ channel (BK or MaxiK) (423). Although this result is intriguing, the possibility that ROMK subunits may constitute the pore-forming components of mitochondrial KATP channels remains to be independently confirmed by other laboratories and with genetic approaches.
2. Kir6.x subunits
A very large number of studies, performed with diverse approaches, have established that the Kir6 subfamily members (Kir6.1 and Kir6.2) are the pore-forming components of the KATP channel (Tables 1 AND 2). The first member of the Kir6 subfamily (Kir6.1; initially named uKATP-1) was identified with homology cloning procedures by screening a rat pancreatic cDNA library with a GIRK1 (Kir3.1) probe (374). Kir6.1 was described to have a ubiquitous tissue expression distribution. The low amino acid identity between Kir6.1 and previously identified Kir subfamily members was responsible for placing this subunit into a new subfamily, Kir6. The second member of the Kir6 subfamily (Kir6.2; initially named BIR or β-cell inward rectifier) was subsequently identified by homology cloning, using Kir6.1 as a probe, from a human genomic library (372). Kir6.2 mRNA expression was found in pancreatic β-cell, heart, skeletal muscle, and brain (372). Topology analysis predicts each of the Kir6.x subfamily members to have two transmembrane regions with intracellular NH2 and COOH termini (Figure 4). There is a 71% amino acid identity (85% similarity) between human Kir6.1 (Genbank accession number NP_004973) and human Kir6.2 (NP_000516). The amino acid differences occur mostly in the NH2 and COOH termini and in an extracellular region (between M1 and M2), where Kir6.1 has a 9-amino acid insert in this region (Figure 6). Kir6.1 and Kir6.2 are well-conserved across mammalian species. Although a third subfamily member has been identified in Zebrafish (917), the Kir6 subfamily appears to be restricted to two members in mammals.
C. Regulatory Subunits: The Sulfonylurea Receptors
Kir6 subunits are unique among the K+ channel subunits since the presence of an auxiliary sulfonylurea receptor (SUR) subunit is an absolute requirement to form a functional channel. SUR1 was first identified by photolabeling with glibenclamide (an antidiabetic sulfonylurea), by degenerate PCR, and by screening rat insulinoma (RINm5F) and hamster insulin-secreting tumor cell (HIT T15) cDNA libraries (4). Northern blot analysis showed high SUR1 mRNA expression in pancreas, brain, and heart. SUR2A (one of the major splice variants of the ABCC9 gene) was identified by homology screening of rat brain and heart cDNA libraries using SUR1 as a probe (371). A second major splice variant (SUR2B) was identified by differential display PCR and RACE amplification of mouse heart cDNA libraries (117). SUR2A mRNA levels were reported to be high in heart, skeletal muscle, and ovary and at moderate levels in brain, tongue, and pancreatic islets (although not expressed in insulin-secreting cell lines), whereas SUR2B was expressed ubiquitously (117, 371). It should be noted that other splice variants exist for both SUR1 and SUR2 (114, 315, 727), but these have not been examined in detail and the published literature focuses largely on full-length SUR1, SUR2A, and SUR2B.
SUR proteins belong to superfamily ATP-binding cassette (ABC) proteins (Figure 3B), and they are established as being subunits of the sarcolemmal/plasmalemmal KATP channel protein complex (Figures 4 AND 5). Most of the ABC proteins transport various molecules across cell membranes at the expense of ATP hydrolysis. A defining characteristic of these proteins is the presence of intracellular ATP-binding domain(s), known as nucleotide-binding folds (NBFs) that contain characteristic Walker A and B motifs, separated by ∼90–120 amino acids (168). There are seven mammalian ABC gene subfamilies. SURx proteins belong to the ABCC subfamily, which also contains the CFTR (or ABCC7). SURx proteins have 17 transmembrane regions, arranged in three domains: TMD0, TMD1, and TMD2, respectively, consisting of transmembrane segments 1–5, 6–11, and 12–17 (Figures 3B AND 4). This topology, originally proposed by sequence comparison with multidrug resistance-associated proteins (825), was confirmed experimentally (134).
As mentioned above, there are two major SUR2 splice variants: SUR2A (371) and SUR2B (382). The two variants differ from each other in their distal COOH-terminal 42-amino acid residues (382) due to alternative exon usage. This rather small difference gives rise to significant functional diversity. For example, pinacidil activates SUR2A/Kir6.2 and SUR2B/Kir6.2 KATP channels with similar potency (Table 3), whereas nicorandil activates the SUR2B/Kir6.2 channel more than 100 times more potently than SUR2A/Kir6.2 channels (731). Kir6.2/SUR2B channels are more efficiently activated by cytosolic MgADP than Kir6.2/SUR2A channels (546, 660). Moreover, whereas Kir6.2/SUR2B channels are strongly activated by KATP channel openers such as diazoxide under basal conditions, there is a requirement for elevated cytosolic MgADP levels in order for diazoxide to activate Kir6.2/SUR2A KATP channels (546, 660, 889). The tissue distributions of SUR2A and SUR2B are also distinct. SUR2A is expressed at high levels in cardiac ventricle [but not in the rodent atrium (18)], skeletal muscle, and ovary and at moderate levels in brain neurons, tongue, and pancreatic islets (18, 371, 932); SUR2B is more widely expressed, for example, in vascular smooth muscle, heart, cardiac specialized conduction system myocytes, vascular endothelium, lung epithelium, hair follicles, renal proximal tubules, renal tubular epithelial cells, microglia, astrocytes, and the dentate gyrus (48, 50, 86, 144, 439, 491, 496, 626, 637, 705, 734, 906, 907, 932, 933). The transcriptional mechanisms responsible for the differential tissue distributions of SUR2A and SUR2B remain to be elucidated.
1. Nucleotide binding domains
SURx proteins have two NBFs; NBF1 is located between TMD1 and TMD2, whereas NBF2 is COOH terminal to TMD2 (Figure 3B). Walker A and Walker B motifs, associated with nucleotide hydrolysis, are present in each of the NBFs and are typical nucleotide binding domains. The consensus sequence for a Walker A motif is G-X-X-X-X-G-K-[TS]. The highly conserved lysine (K) residue forms hydrogen bonds with the oxygen atoms of the α- and γ-phosphates of the bound nucleotide (844). The aspartate residue in the Walker B motifs (the consensus sequence is [RK]-4X-G-4X-L-4Φ-D) coordinates Mg2+ binding. Mutagenesis of the Walker A lysine residues in (e.g., K719 and K1384, respectively, in NBF1 and NBF2 of SUR1) or the Walker B asparate residues (e.g., D854 of SUR1) disrupts certain aspects of KATP channel function, such as the activation by MgADP and KATP channel opening compounds (14, 285, 827).
D. Subunit Composition Determines the KATP Channel′s Biophysical and Pharmacological Properties
The diversity of KATP channel subtypes in various cells originates in part by specific combinations of the various Kir6.x and SURx subunits (Table 2). For example, ventricular sarcolemmal KATP channels and those in pancreatic β-cells have the same pore-forming subunit (Kir6.2) but differ in the use of the regulatory subunit (SUR2A in the case of the cardiac channel and SUR1 in the pancreas). Atrial KATP channels (at least in the rodent) appear to rely on SUR1 subunits. Smooth muscle KATP channels, in contrast, appear to be composed of Kir6.1 subunits in combination with SUR2B subunits, and endothelial KATP channels may be composed of heteromeric Kir6.1/Kir6.2 pore-forming subunits in combination with regulatory SUR2B subunits. The subunit combination of mitochondrial KATP channels has not been resolved, but roles for Kir1.1 and/or SUR2 short-form splice variants have been proposed (Figure 5).
E. Interactions of KATP Channel Subunits With Other Subunits and Proteins
A common biological theme is that the biological functions of proteins, including those of ion channels and transporters, are regulated by other subunits of the protein complex. The associated subunits may be responsible for regulating channel function and intracellular trafficking, such as the Kvβ subunit that promotes inactivation and surface expression of Kv channels (126), or the early formation of a Kir6.x/SURx complex, which overcomes the ability of ER-resident proteins to prevent surface expression (see sect. V for more details).
1. Metabolic enzymes
It has long been recognized that cardiac KATP channels are preferentially regulated by glycolytically derived ATP. With experiments performed in the open-cell and inside-out patch-clamp configurations, the Weiss laboratory has demonstrated that cardiac KATP channels are closed equally well by ATP production from oxidative phosphorylation, the creatine kinase shuttle system, and glycolysis (866). However, when intracellular ATP consumption was stimulated, glycolytic enzymes were far more efficient in blocking KATP channel activity. This result suggested that the KATP channel activity is regulated by glycolysis, which alters the nucleotide concentrations in a small submembrane compartment in the immediate microenvironment of the KATP channel complex. Preferential regulation of KATP channels by glycolytically derived nucleotides has also been noted in other tissues, including enterocytes (191) and certain types of neurons (529, 530, 902). Findings such as these have been interpreted that key glycolytic enzymes are located near KATP channels at the plasma(sarco)lemma (866). Indeed, using two-hybrid and coimmunoprecipitation assays, the glycolytic enzymes aldolase (ALD), glyceraldehyde 3-phosphate dehydrogenase (GAPDH), triosephosphate isomerase (TPI), pyruvate kinase (PKM), and lactate dehydrogenase (LDH) were found to be de facto interacting proteins of the cardiac KATP channel (139, 179, 357, 409). At least one of these enzymes (aldolase) appears to interact with an intracellular coiled-coil domain within the NBF1, close to the 12th transmembrane domain of SUR2 (357). Using unbiased proteomic approaches in a search for KATP channel interacting proteins, we recently performed mass spectrometry of immunoprecipitates of KATP channel complexes (357, 433). Pathway analysis of the identified interacting proteins revealed glycolysis to be the most significantly represented pathway (357, 433). Patch-clamp experiments showed that the catalytic activity of glycolytic enzymes can regulate KATP channel activity, even when experiments are performed in with inside-out membrane patches (179). The distal, ATP-producing glycolytic enzymes, such as PKM were particularly effective in regulating KATP channel activity (357). Experimentally, KATP channels readily open when inhibiting mitochondrial ATP production (e.g., with cyanide or DNP) and the physiological relevance of glycolytic regulation of KATP channels is not immediately obvious. Although it remains an untested idea, it is possible that local ATP depletion from glycolytic sources is responsible for cardiac KATP channel opening and action potential duration adaptation that occurs with a sudden increase in the heart rate (941). This mode of regulation may become more pronounced with aging, since glycolytic inhibition in aged rat hearts causes action potential shortening and arrhythmias that can be prevented by KATP channel block with glibenclamide (575). Competition for local glycolytically derived ATP may also be responsible for the known functional interaction between the KATP channel and the Na+/K+ pump, which has been observed in several tissues, including frog skin (829), neurons (8), the kidney (548), skeletal muscle (663), smooth muscle (269), and the heart (411, 646). Although the physiological relevance of this functional interaction is obvious, it has been suggested to be important in the protective effects of ischemic preconditioning (IPC) (331).
Aside from glycolytic enzymes, KATP channels also interact with (and are regulated by) enzymes of intracellular phosphotransfer networks (197), which transfer phosphoryls between different intracellular compartments without causing significant alterations in the cytosolic levels of adenine nucleotides (914). These phospho-relays provide energy transfer pathways between flux-generating (ATPases) and flux-responding (glycolysis and oxidative phosphorylation) processes. Adenylate kinase (AK), for example, catalyzes the reversible phosphotransfer between ADP and ATP in the presence of AMP and Mg2+ as follows:
It has been suggested that AMP, which is produced by other biochemical pathways, is delivered to the membrane by the AK-catalyzed phosphotransfer system. In the microenvironment of the channel complex, AK then converts the AMP to ADP (and in the process consumes ATP). The resulting change in the ATP/ADP ratio in the microenvironment of the channel is sufficient to lead to KATP channel opening (100, 208). In contrast, creatine kinase (CK), which also interacts with KATP channel subunits (141), mediates this reversible phosphotransfer reaction:
2. Other interacting proteins
Proteomic analysis has identified several types of proteins as potential KATP channel interacting partners in the cardiovascular system (the heart and endothelial cells) (357, 433). The veracity of the majority of these proteins awaits independent confirmation. It is likely that this list includes many false positives and several proteins that may only interact weakly or transiently. The proteins identified with quantitative mass spectrometry at high confidence (i.e., they were immunoprecipitated by at least two separate antibodies and significantly enriched relative to negative control reactions) included proteins involved in metabolic pathways, cytoskeletal organization (actin, myosins and microtubules), and subcellular protein localization/trafficking. In addition to this list of candidates, there is evidence that syntaxin 1A may interact with SURx subunits and that this interaction regulates the function of KATP channels in pancreatic β-cells and in cardiac myocytes (104, 108, 421, 422, 634) and that this binding may depend on PIP2 levels (883). In cardiac muscle, ankyrin-B was found to interact directly with the Kir6.2 subunit and to regulate the membrane expression and function of KATP channels in the heart (494). The interaction with ankyrin-B is likely to reflect a trafficking or membrane anchoring function, and it is possible that βIV-spectrin may have a similar role, as has been described for the pancreatic KATP channel (449). In vascular smooth muscle, caveolin-1 was described to interact with the KATP channel and that this suppresses gating by reducing the sensitivity of channels to the stimulatory influence of MgADP (162). Kir6.2 was also described to associate with caveolin-3 (but not caveolin-1), and it was found that ventricular KATP channels are negatively regulated by caveolin (256). A possible role for this interaction and the presence of KATP channels in caveolae is to localize receptor signaling to KATP channels in smooth muscle and the endothelium (606, 687, 688). A full understanding of subunits of the KATP channel megadalton protein complex, and how they regulate the function and trafficking of KATP channels in the cardiovascular system, is still in its infancy and much can be learned from further exploring this topic.
IV. MOLECULAR MECHANISMS THAT REGULATE KATP CHANNEL FUNCTION
A. Kir6 Subunits Interact With SUR Subunits: Stoichiometry of Interaction
A novel radioiodinated azidoglibenclamide analog was found to photolabel a ∼175 kDa protein in microsomes isolated from porcine cerebral cortex (709). Interestingly, the same group later found that by omitting the “boiling step” prior to SDS-PAGE, a novel colabeled 38 kDa protein was revealed (710). The Bryan laboratory demonstrated the 38 kDa protein to be Kir6.2, which provided the first biochemical evidence for direct physical interaction between Kir6.2 and SUR1 (122). In the latter study, heterologously expressed Kir6.2 and SUR1 subunits (in Cosm6 cells) were both labeled by 125I-azidoglibenclamide, whereas other Kir subunits (Kir1.1 or Kir3.4) were unlabeled. Moreover, Kir6.2 and an NH2-terminal His-tagged SUR1 could be copurified from digitonin-solubilized membranes of COSm6 cells expressing both subunits, with an estimated molecular mass of the protein complex to be ∼950 Da (122), consistent with at least an octameric composition of four Kir6.2 and four SUR1 subunits. Coimmunoprecipitation assays with anti-Kir6.2 antibodies also demonstrated a specific interaction between Kir6.2 and SUR1 in membranes isolated from COS cells co-expressing the two subunits (522) and between the in vitro translated Kir6.2 and SUR2A proteins (522).
The stoichiometry of Kir6.2 and SUR1 interactions has been further investigated with coexpression approaches. The Seino group found that currents could be recorded from COS1 cells transfected with a concatemeric construct consisting of a tandem Kir6.2-SUR1 subunit (373), suggestive of a 1:1 stoichiometry of assembly. The Bryan group used a similar approach and reached the same conclusion (122). Shyng and Nichols (737) extended this approach by generating a mutant Kir6.2-N160D subunit, which confers strong rectification properties to the normally weakly rectifying channel (see sect. IIE). By coexpressing different ratios of wild-type or Kir6.2-N160D mutant subunits, or dimeric (SUR1-Kir6.2) constructs, they were able to determine that the KATP channel pore is lined by four Kir6.2 subunits and that each Kir6.2 subunit requires one SUR1 subunit to generate a functional channel, a model that is consistent with an octameric or tetradimeric (Kir6.2-SUR1) structure. One assumes that the Kir6.2-SUR2x complex also has this octameric structural organization.
B. The Channel Pore
The pore of a K+ channel is comprised of a conserved pore-forming domain (or the “P loop”) between TM1 and TM2, which is a short amino acid segment that dips into the membrane (188). The P domain of K+ channel principal subunits is critically important for channel function and ion permeation. Mutagenesis and three-dimensional structural studies have demonstrated that this domain forms the K+-selective pore of the channel (188, 341, 775). The primary sequence of the P loop of Kv subunits has the signature sequence T-V-G-Y-G (standard single letter amino acid codes) (126). For the Kir subunits, the P-loop “signature sequence” is represented by T-I-G-[YF]-G (Figure 6).
C. Block of the KATP Channel by Intracellular ATP
The normal requirement is for Kir6.x and SURx subunits to assemble before surface expression occurs. However, experimentally it is possible for channels to be formed by Kir6.2 subunits in the absence of SURx. The Kir6.2 COOH terminus contains an ER retention signal (see sect. VB). Mutagenesis of this signal sequence, or deleting it with a COOH-terminal truncation (e.g., Kir6.2ΔC26), allows SURx-independent Kir6.2 surface expression. These Kir6.2ΔC26 channels can still be blocked by ATP (821), demonstrating that the Kir6.2 subunit has an intrinsic ATP inhibitory site. However, no classical consensus motif for ATP binding is evident in the amino acid sequence of Kir6.2. Neutralization of a specific lysine residue Kir6.2ΔC26-K185Q was found to substantially decrease ATP sensitivity (821), suggesting that this lysine might be involved in ATP-binding. Systematic mutagenesis of “intracellular” charged residues identified several amino acids that alter ATP sensitivity (e.g., R50, C166, I167, T171, K185) (820). Some of these mutations lowered the intrinsic open probability of the channel. The apparent change in ATP sensitivity might have been due to the fact that ATP binding is favored during the long closed state of the channel (12, 739). Mutagenesis of R50 and K185 markedly altered ATP sensitivity without altering kinetics and were therefore assumed to be involved in ATP binding (820). R201 has been identified as another residue critical for ATP binding in subsequent mutagenesis studies (667, 738). It is believed that channel inhibition by ATP may involve interaction of ATP′s α-phosphate with R201, the β-phosphate with K185, and the γ-phosphate with R50 from an adjacent subunit (405, 667) (Figure 7). A more complete structural model of the ATP-binding site of the Kir6.2 subunit has been published (16). The KATP channel is powerfully activated by phosphatidylinositol 4,5-bisphosphate (PIP2) in the membrane (see below). The Kir6.2 residues R54, R176, R177, and R206 are thought to bind PIP2 (52, 738, 740). The PIP2 binding site overlaps with that of ATP, and it is believed that ATP binding to Kir6.2 destabilizes the channel′s interaction with PIP2, thereby contributing to a transition to a closed state (405, 596, 740).
D. Simulation of KATP Channel Opening by Intracellular MgADP
ATP blocks the KATP channel by binding to the Kir6.2 subunit (see below). However, coexpression of Kir6.2 with SURx strongly alters the channel′s ATP sensitivity. For example, ATP inhibits heterologously expressed Kir6.2/SUR1 currents with a half-maximal inhibitory concentration (IC50) of ∼10–20 μM (596), whereas channels consisting of Kir6.2 alone (e.g., Kir6.2Δ26) have a 10-fold reduced ATP sensitivity (IC50 ∼140 μM) (37, 821). Binding of nucleotides (and Mg2+) to either NBF1 or NBF2 could therefore potentially participate in ATP inhibition. However, mutagenesis within NBF1, NBF2, or the linker region between the NBFs, does not prevent ATP inhibition (285, 739). It is likely, therefore, that the altered ATP sensitivity of coassembled Kir6.2/SURx channels is not due to ATP binding to the NBFs. Recent studies suggest that interaction domains between Kir6.2/SURx, mediated by cytoplasmic domain interfaces near the membrane, may regulate ATP sensitivity (644). However, the molecular underpinnings by which Kir6x/SURx assembly affects ATP sensitivity remain elusive.
Nucleotides do indeed bind to SURx′s NBF1 and NBF2, and the reader is referred to excellent reviews on SUR proteins and their nucleotide binding properties (5, 25, 92, 543, 719). Support for the notion that MgADP stimulates the KATP channel by interacting with SURx comes from the finding that Kir6.2 expressed in the absence of SURx (e.g., Kir6.2Δ36) is not activated by MgADP (287). Both MgATP and MgADP have been reported to activate KATP channels by acting on the NBFs. For example, mutagenesis of the conserved Walker A lysine residues in (e.g., K719 and K1384 in SUR1) or the Walker B asparate residues (e.g., D854 of SUR1) inhibits activation of the KATP channel by MgADP (14, 285, 827). Direct evidence for nucleotide binding to NBFs come from photoaffinity labeling with 8-azido-[32P]ATP (827). Mutagenesis experiments revealed that SUR1 NBF1 has high affinity for ATP binding (Mg2+ is not required for ATP binding). Moreover, they showed that MgADP binding to NBF2 antagonized ATP binding to NBF1. These findings were verified when the NBF1 and NBF2 domains were expressed individually. NBF1 was affinity labeled by 8-azido-[32P]ATP with the radioactive phosphate group present in either the α or γ position, whereas NBF2 was labeled only by 8-azido-[α-32P]ATP (544, 545). Direct evidence for ATP hydrolysis was presented with the finding that immunoprecipitates obtained with anti-KATP channel subunit antibodies contain ATPase activity (66). Moreover, the ATPase activity of purified SUR2 NBF2 was significantly higher than that of NBF1 (66, 939). Support for the idea that ATP hydrolysis is required for KATP channel activation comes from experiments using nucleotide trapping procedures (vanadate or beryllium-fluoride) to stabilize the ATPase cycle in distinct conformations (939). It has also been demonstrated that disease-causing mutations in SUR1 NBF2 (R1380L and R1380C, which are linked to neonatal diabetes) activates the KATP channel by increasing the channel complex′s ATPase activity (167). Collectively, these data suggest that ATP bound to NBF2 is hydrolyzed to MgADP and that NBF1 has little ATPase activity. Although elegant, the experiments with isolated NBF proteins should be interpreted with some caution since in general NBF dimerization is favored upon nucleotide binding and ATP hydrolysis occurs with cooperative interaction of the dimerized NBFs. Indeed, the ATPase activity of SUR1 is substantially less than that of the isolated NBF2 (166). Furthermore, not all data are consistent with the idea that ATP hydrolysis is required for basal KATP channel function. For example, mutagenesis of the conserved Walker A lysine and/or Walker B aspartate residues (see sect. IIIC1), which as expected disrupts ATPase activity in other ABC proteins (777, 795), does not substantially affect basal KATP channel function or ATP inhibition (66, 151, 285). It may be possible for the hydrolyzed MgADP to be trapped in NBF2, which may be responsible for channel activation. This view is not entirely consistent with the strong activation of KATP channels by MgADP in excised patches (at a constant MgATP concentration). An alternate view is that ATP hydrolysis at NBF2 is epiphenomenon and that MgADP binding directly activates the channel (596). Further research is needed to resolve these issues.
E. Mechanisms of Rectification
As mentioned in section II, the current-voltage relationship of KATP channels exhibit weak inward rectification properties (reduced outward current at voltages positive of the K+ equilibrium potential). Physiologically, inward rectification of K+ channels is an important energy-saving mechanism: the K+ efflux resulting from open K+ channels during the long cardiac action potential must be restored by the Na+/K+ pump at the expense of ATP hydrolysis. A low overall membrane conductance during the long action potentials and limiting outward K+ flux by inward rectification mechanisms is therefore energetically beneficial. Early studies have noted a negative slope region of the KATP channel unitary current-voltage relationship (607, 815), indicative of inward rectification. It was found that “intracellular” Na+, TEA+, and Tris can cause inward rectification; intracellular pore block (to reduce outward current) was proposed as the mechanism (135, 416). Intracellular Mg2+, at concentrations in the low millimolar range, was subsequently demonstrated unidirectionally to block outward KATP channel current (359). It is now recognized that physiological levels of intracellular Na+ and Mg2+ contribute to KATP channel inward rectification. Naturally occurring intracellular polyamines (e.g., spermine, spermidine, and putrescine) are also major determinants of KATP channel current rectification (225, 521, 597). These long and thin polyvalent cations enter the KATP channel subunit inner vestibule and interact with residues of the Kir6.2 subunit to block outward K+ flux (524, 736). As demonstrated by a Kir3.1 crystal structure (603), it is likely that polyamines bind within an extended cytoplasmic pore that is formed by the Kir6.2 intracellular NH2 and COOH termini (470).
F. Regulation of the KATP Channel by PIP2
Phospholipids, such as the membrane PIP2, activates the KATP channel (52, 214, 349, 740) and increases the channel′s intrinsic open probability (Po) by prolonging the mean open time and shortening the mean closed time (214). The sensitivity of the KATP channel to ATP inhibition is much reduced by PIP2 (52, 214, 740). A part of this effect is indirect (210, 814) since ATP stabilizes the closed state of the channel (736, 814), which is less accessible when the Po is high. The high Po may also be responsible for the effect of PIP2 to mitigate activation of KATP channels by MgADP, to reduce channel inhibition by sulphonylureas (463), and to attenuate rundown (214). However, the ATP sensitivity of a COOH-terminal truncated Kir6.2 channel is also reduced by PIP2, and this occurs in the absence of significant effects on Po (215), which suggests that PIP2 additionally have direct effects on ATP binding. Indeed, competition occurs between ATP and PIP2 binding for biochemically purified Kir6.2 protein (857) or of the Kir6.2-containing channel (532). Molecular modeling and mutagenesis studies suggested that the ATP and PIP2 binding sites within the Kir6.2 subunit overlap (145, 307). Thus binding of PIP2 in the membrane to the Kir6.2 subunit seems to have important functional consequences on KATP channel function. An open question remains whether this mode of channel regulation occurs physiologically. Using a GFP-tagged PLCγ PH-domain construct to visualize intracellular PIP2, a case has been made that α1-adrenoceptor stimulation inhibits ventricular myocyte KATP channels by reducing the membrane PIP2 levels (332). There are not many other examples of studies showing that PIP2 regulation of KATP channels occur under physiological, pharmacological, or pathophysiological conditions. PIP2 may also have a function during biosynthesis and vesicle trafficking of ion transporters and channels: low levels of PIP2 in the intracellular membranes may keep channels inoperative until they reach the surface membranes to be activated by higher local PIP2 levels (350).
V. SYNTHESIS, TRAFFICKING, AND DEGRADATION OF KATP CHANNEL PROTEINS
It is easy to visualize KATP channels as being uniformly expressed on the surface of cardiac myocytes at a constant density. However, the surface expression of KATP channels may be quite labile and regulated by a fine balance between de novo synthesis, assembly, anterograde trafficking, subunit interaction, posttranslational modifications such as glycosylation, membrane anchoring, and endocytic recycling (Figure 8). Please refer to an excellent recent review that summarizes the literature and our current understanding of the dynamic regulation of ion channels in cardiomyocytes (45). The text below specifically deals with KATP channels, which are not highlighted in the review mentioned. Much of the text that follows is based on data obtained in heterologous expression systems or in cell types other than cardiomyocytes. Given the evolutionary conservation of trafficking mechanisms from yeast to mammals, this information is likely directly relevant to the cardiomyocyte, smooth muscle cell, or endothelial cells.
A. Regulation of KATP Channel Subunit Expression
The origin of any protein is the nucleus, and the transcriptional regulation of gene expression constitutes an important control point. Transcriptional regulation of the KATP channel subunit genes (KCNJ8, KCNJ11, ABCC8, and ABCC9) has not been studied systematically. A recent study showed that the ABCC8 and ABCC9 promoters can be methylated in HL-1 cells and that methylation decreases mRNA expression (220). Although these genes were not methylated in the mouse heart, the possibility is raised that DNA methylation may regulate transcriptional activity of KATP channel genes.
The transcription of KATP channel subunit mRNA can be regulated during pathophysiological situations such as ischemia/reperfusion. In a model of left coronary occlusion rat hearts, ion channel remodeling was demonstrated to occur with the development of postischemic dilated cardiomyopathy (381). Specifically, Kir6.1, SUR1, and SUR2A mRNAs were upregulated 8–20 wk postinfarction. A strong increase in Kir6.1 mRNA expression was observed after renal ischemia or regional myocardial ischemia in the rat (9, 724), without significant changes of Kir6.2 and SUR2 mRNA expression levels. The local renin-angiotensin system and tumor necrosis factor (TNF)-α may play a role in the upregulation of Kir6.1 after ischemia (10). Kir6.1, Kir6.2, SUR1, SUR2A, SUR2B mRNA expression is strongly upregulated in the border zone of the rat heart at 7–140 days after coronary artery ligation, and their increased expression levels correlate with that of forkhead transcription factors FoxF2, FoxO1, and FoxO3 (638). Gel-shift assays demonstrated that FoxO1 binds to Kir6.1 in mouse atrial HL-1 myocytes (638).
During chronic hypoxia, upregulation of Kir6.1 protein levels occurs in rat heart and brain, whereas Kir6.2 is downregulated (253). Hypoxia was also reported to upregulate Kir6.1 (560) and SUR2A (140) mRNA in H9c2 rat heart cells. The upregulation of SUR2A mRNA expression by chronic hypoxia was described to occur independently of HIF-1α activity, independent in the H9c2 cells (140). Kir6.1 is also upregulated in the atria of children with congenital heart disease, possibly due to venous hypoxemia; this effect appears to be mediated by hypoxia-induced activation of HIF-1α and FOXO1 (652). In patients with mitral valve disease, cardiac Kir6.2 mRNA expression positively correlates with venous oxygen content and with HIF-1α mRNA expression, further supporting a role for hypoxia-sensitive transcription factors to regulate KATP channel expression. The significance of strong hypoxia-induced upregulation of Kir6.1 in the heart is not obvious at present, since this subunit is generally not assigned a function in cardiomyocytes. Further studies are needed to determine if Kir6.1 has a novel functional role in posthypoxic ventricular myocytes or whether Kir6.1 is upregulated in other cell types (e.g., in the endothelium or coronary arteries, perhaps as a result of increased vascularization).
B. Protein Assembly and Quality Control Mechanisms in the ER
Membrane proteins, such as ion channel subunits, are translated in the rough ER and then enter the endoplasmic reticulum in an unfolded state through the translocon, which is a protein-conducting channel (656). Within the ER, many proteins coordinate to ensure the proper function and assembly of newly synthesized proteins. Chaperone proteins assist with folding and membrane integration, whereas other modifying enzymes ensure appropriate posttranslational modifications, such as glycosylation or disulfide bond formation (81). This is also the time when subunits are assembled into their unique multisubunit complexes. It is only when the channel protein complex is correctly assembled and modified that it leaves the ER to be transported through the anterograde trafficking pathway and the Golgi apparatus, with its final destination as the surface membrane.
Not all proteins are folded properly, despite the existence of a multitude of control mechanisms. If misfolded proteins accumulate, an ER stress response may ensue (855). The misfolded proteins are normally retained within the ER and are disposed of by the ERAD system (673). Upon recognition, a misfolded protein is transported from the ER to the cytoplasm (also known as retrotranslocation), where it is ubiquitinated by E3 ligase and released in the cytoplasm for degradation in a nonlysosomal manner by the proteasome. There is evidence that KATP channels can be degraded by the ERAD pathway. First, SUR1 and Kir6.2 have been demonstrated to form a complex with Derlin-1 and an associated p97 protein (858). Derlin-1 participates in the recognition of misfolded proteins, and it forms the retro-translocation channel, whereas p97 is an ATPase that links the misfolded proteins to ubiquitin ligases (656). Second, KATP channel subunits can be ubiquitinated (788). Third, KATP channel subunit degradation can be prevented by MG132, which is an inhibitor of the proteasome (788, 891). It should be noted that these studies were performed with the Kir6.2/SUR1 channel (pancreatic β-cell form). It remains to be demonstrated that KATP channels in the cardiovascular system participates in ERAD.
Correctly folded proteins are assembled within the ER before exiting and being transported to the surface. Following the molecular cloning of KATP channel subunits, an early observation was that Kir6.x and SURx subunits need to be coexpressed to generate functional channels (371, 374). A key finding in understanding this phenomenon came from a study in which a Kir6.2 subunit with a truncated COOH terminus (by 26–36 amino acids) was found to express functional channels in the absence of SURx subunits (821). We now know that trafficking of unassembled Kir6.x or SURx subunits to the plasma membrane is prevented by an arginine-based ER-retention/retrieval signal, composed of the three amino acids RKR, respectively, in the Kir6.2 COOH terminus and in a cytoplasmic loop of SURx located between the 11th transmembrane domain and NBF1 (Figure 4) (916). It should be noted that the arginine-based motif is “RKR” in SUR1 and “RKQ” in SUR2. The dibasic motif RKR may interact with COPI proteins, which results in ER retention (610, 910). Subunit assembly is thought to shield these dibasic motifs from ER resident proteins. In addition to the ER retention signal present in the distal Kir6.2 COOH terminus, other signals within the M2 transmembrane and the proximal COOH-terminal regions also contribute to the inability of Kir6.2 to traffic to the membrane in the absence of SURx (361). The presence of these motifs therefore most likely ensures that only fully assembled octametric KATP channel complexes are able to continue on trafficking route towards the plasma membrane (915).
Do KATP channel subunits have anterograde trafficking signals? Certain inward rectifier K+ channel subunits, such as Kir2.1, efficiently traffic to the surface membrane. In a study with subunit truncation and chimeras with other proteins, the presence of an ER export signal (FCYENE) has been identified in the Kir2.1 COOH terminus (531). The KATP channel pore-forming Kir6.x subunits lack a corresponding motif. An anterograde trafficking signal that is required for the assembled KATP channels to exit the ER/cis-Golgi compartment was described for the SUR1 subunit (725). Truncation experiments suggested that the last 5–13 amino acids of the SUR1 COOH terminus are required for surface expression. At present, this study is unconfirmed, given that other groups found that SUR1 COOH-terminal truncations did not markedly inhibit surface expression (268, 685, 712). Thus it is therefore questionable whether KATP channel subunits contain any specific anterograde trafficking signals. It is known, however, that binding of the channel complex to 14-3-3 proteins promotes the cell surface transport of correctly assembled complexes and that this interaction is without effect on the KATP channel function (346).
C. Anterograde Trafficking of the Assembled KATP Channel Protein Complex
After exiting the ER, KATP channel proteins presumably follow the regular anterograde trafficking pathway through the Golgi apparatus to the cell surface (Figure 8). Protein glycosylation occurs in the ER and Golgi compartments (666). SUR1, for example, is modified by this form of posttranslational modification at two N-linked glycosylation sites (651). This mature, complex glycosylated form of SUR1 resolves with lower electrophoretic mobility when studied with SDS-PAGE and can be distinguished from the core-glycosylated form (916). The N-glycosylation that occurs in the ER is sensitive to Endo H digestion, whereas proteins glycosylated in the Golgi resist the Endo H treatment. Since the mature form of SUR1 is resistant to Endo H digestion (18, 725), we can conclude that KATP channels traffic through the Golgi, where they are glycosylated. Although not demonstrated, one can assume that KATP channels follow the canonical forward trafficking pathway (749). Packed in vesicles, they are transported through the cis-Golgi and trans-Golgi along microtubules tracks to the cell surface, driven by specific molecular motors.
D. Membrane Delivery and Anchoring
The delivery of new channels to the membrane, as well as their anchoring at the membrane, is a topic recently covered in a review (45). The fusion of trafficking vesicles to the membrane is controlled by the SNARE complex (750). Proteins of the SNARE complex have been described to interact with ion channel subunits. For example, SNAP-25 has been described to interact with several types of Ca2+ channel subunits and the K+ channels subunit Kv2.1 (335, 541, 875). Another SNARE protein, syntaxin-1A, also interacts with Ca2+ and K+ channel subunits and functionally regulates these channels (119, 174, 541, 562, 819). Syntaxin-1A also interacts with proteins of the ABCCx gene family, including CFTR (ABCC7), SUR1 (ABCC8), and SUR2 (ABCC9) (252, 421, 634). Syntaxin-1A regulates both the trafficking and gating of CTFR (791). By binding to the intracellular nucleotide binding folds of SUR1 and SUR2, syntaxin-1A negatively regulates KATP channels (104, 421, 634). The inhibition is due in part by accelerated endocytosis of surface channels, but it also regulates KATP channel surface expression by inhibiting the early secretory pathway (108).
Scaffolding proteins have long been recognized to localize ion channels and to anchor them to the membrane in specific areas, such as the postsynaptic density (PSD) of neurons (846). The membrane-associated guanylate kinase (MAGUK) protein family has a central role in controlling the localization of proteins in the PSD. Members of the MAGUK protein family, including SAP97 and PSD-95, are known regulators of ion channel trafficking and localization (242), but there are no reports indicating that KATP channels are regulated by these proteins. In contrast, elements of the cytoskeletal network have a recognized role. Ankyrin functions as a linker between integral membrane proteins and the actin-spectrin-based cytoskeleton (146, 672). Disruption of any element within this cortical cytoskeletal network has effects on KATP channel function. Disrupting the actin network (e.g., with cytochalasin B) enhances KATP channel activity by reducing its sensitivity to inhibitory intracellular ATP (802) and decreases channel block by glibenclamide (82). βIV-Spectrin has been shown to serve as a structural and signaling platform for KATP channels in the pancreatic islet (449), and one assumes that this is also the case in the cardiovascular system. Ankyrin-B (but not ankyrin-G) interacts with Kir6.2 (but not Kir6.1) at a motif contained within residues 313–325 of the subunit′s COOH terminus (448). Disrupting ankyrin-B expression impairs KATP channel KATP channel surface expression, but also has functional effects that have been reported. For example, channels comprised of Kir6.2 subunits that lack the ankyrin-B binding domain have a decreased ATP sensitivity (448, 494). Thus scaffolding and anchoring proteins can regulate both the surface density and function of KATP channels, and this mode of regulation can be important in human disease and cardiomyopathies (148).
After reaching the surface membrane, KATP channels do not remain there. In general, cells internalize the equivalent of their entire surface area up to five times per hour, and most endocytosed proteins and lipids are recycled back to the plasma membrane (760). Some membrane proteins can be reused hundreds of times through specific recycling pathways (282). Endocytic recycling is a rapid process, and it occurs in the order of minutes (760). For ion channels, this process is potentially a powerful mechanism to control their surface density and hence their function. The surface density is determined by a delicate balance between the rates of endocytosis and exocytosis, which in turn are regulated by specific specialized trafficking proteins. We are only now starting to understand the complex events involved in the endocytic recycling of KATP channel subunits.
Several reports have demonstrated that KATP channels are endocytosed (89, 363, 537). A dominant-negative dynamin (K44E mutant) was shown to prevent internalization of both SUR1 and SUR2-contaning channels (363, 537). Moreover, KATP channel subunit endocytosis can be inhibited by the dynamin inhibitor dynasore (89) or by a dominant-negative dynamin peptide (741). Dynamin-mediated endocytosis can occur through a clathrin-mediated pathway or may be clathrin-independent, in which case a role may exist for caveolae and/or RhoA (550). Published data are not uniform regarding the endocytosis mechanism(s) of KATP channel subunits. Several studies suggested that KATP channel endocytosis occurs as clathrin-coated pits. For example, internalized channels colocalize with clathrin, and a dominant-negative form of μ2 subunit of the AP2 adaptor complex (D176A/W421A), but not the wild-type μ2, inhibits KATP channel endocytosis (537). Other studies report that, in smooth muscle, PKC-stimulated KATP channel endocytosis is caveolin-dependent since it was inhibited by interventions that deplete membrane cholesterol and siRNA knockdown of caveolin-1 (401). Caveolins may link KATP channels to caveolae and receptor signaling (see sect. IIIE), and this finding does not necessarily signify a role in endocytosis. Moreover, caveolae are relatively immobile membrane structures that are generally not associated with endocytosis (804). Overall, therefore, it appears that KATP channels are constitutively endocytosed through a pathway mediated by clathrin and dynamin, but that there may also be a role for a caveolin-based receptor-stimulated endocytosis (at least in smooth muscle).
There is no agreement regarding the endocytic motifs required for endocytosis. Tyrosine-based motifs YXXØ, where Ø denotes a hydrophobic amino acid, are often found in membrane cargo and serve as a binding motif for the μ2 subunit of the AP2 adaptor complex, which is an early step in clathrin-mediated endocytosis (781). Kir6.2 contains two such motifs: 258YhvI261 and 330YskF333. In one study, mutating the first of these motifs (Y258A) had no effect on endocytosis or surface expression, whereas mutating the second (Y330A) prevented endocytosis (537). In another study, mutagenesis of this motif had no effect on KATP channel endocytosis (89). The reasons for these discrepant findings are unclear, and further studies are needed to determine whether the tyrosine-based motifs in Kir6.2 are relevant. Moreover, SUR2 contains several tyrosine-based motifs; none of these has been studied to date. Dileucine motifs are also involved in endocytosis and intracellular targeting. The [D/E]XXXL[L/I] motif in cargo proteins binds mainly to the σ2 subunit of the AP2 adaptor complex, although variations of this motif can occur (459). Kir6.2 has one such motif (352DhsLL356), and SUR2 has none. In one study, mutagenesis of the lysines at positions 355–356 prevented endocytosis of Kir6.2, either when expressed with SUR1 or SUR2 (363). In another study, the same mutation had no effect on KATP channel endocytosis, but it did increase surface expression, possibly as a result of enhanced recycling (537). The concept that KATP channel endocytosis is solely dependent on Kir6.2 (and embedded signals) is likely too simplistic and has recently been challenged by the finding that SUR1 undergoes endocytosis independent of Kir6.2 (89). Clearly, much remains to be done before a better understanding is reached regarding the mechanisms and signals involved in KATP channel endocytosis.
F. Endocytic Recycling
Following endocytosis, the majority of internalized cargo recycles back to the surface membrane (Figure 8) (549). Experiments with antibody capture assays have demonstrated that KATP channels can be recycled in this manner (538). Endocytic recycling of KATP channels represents a potentially powerful mechanism for regulating their surface density and availability for protection against stress events. Many proteins pass through early endosomes in the recycling process. KATP channels likely follow this route, as evidenced by the partial colocalization of internalized Kir6.2 in pancreatic β-cells with the early endosome antigen 1 (EEA1) protein, which serves as a marker for this cellular compartment (632). The term early endosome actually refers to two types of endosomes, namely, the sorting endosome and the endocytic recycling compartment (ERC) (549). Cargo can be directed from the sorting endosome directly and rapidly (<2 min) to the surface membrane, to late endosomes and/or to the long-lived organelle, the ERC. KATP channel recycling occurs at a time scale of 10–15 min (538), which argues against direct recycling to the membrane. EHD1 (also named Rme1) is a trafficking protein that controls several recycling steps, including the exit of cargo from the ERC to the surface membrane (590). A dominant negative EHD1 construct (EHD1-G429R) was reported in a single study to slow the recycling of the transferrin receptor, but not of KATP channels, suggesting that KATP channels do not utilize this slow recycling pathway (363). There may be a role for late endosomes in KATP channel recycling. At steady state, KATP channels containing SUR1 are mainly expressed on the cell surface, whereas many SUR2-containing channels are localized to intracellular vesicular compartments that are positive for the late endosomal/lysosomal marker, LAMP2 (47). Since the half-lives of SUR1 and SUR2 channels are similar, the SUR2-containing KATP channels are most likely not directed to this pathway for degradation (47). It is known that recycling to the membrane is possible through late endosomal and lysosomal trafficking pathways (654), but it is not yet clear whether KATP channels are directed via this pathway. More experimentation is needed to determine the mechanisms by which KATP channels traffic through the endocytic recycling pathway. This process may be more complex than anticipated, as suggested by a recent finding that chamber-specific subcellular mechanisms may exist in cardiac myocytes (18).
G. Spatial Distribution of KATP Channels in Ventricular Myocytes
The ventricular myocyte is a highly polarized cell. Many channels and ion transporters are localized specifically to the transverse (t)-tubules (749). Experiments with scanning ion conductance microscopy have demonstrated that KATP channels are concentrated at the Z-grooves, which are the sarcolemma regions from which the t-tubules emanate (457). From this observation, clustering of sarcolemmal KATP channels is suggested, with a higher density of channels at the mouth of the t-tubule compared with the lateral membrane. This suggestion is bolstered by the punctate staining pattern in cardiomyocytes when performing immunocytochemistry with antibodies against KATP channel subunits (579). Polarized expression in a ventricular myocyte is also evident at the end-to-end connection between cardiomyocytes (or the intercalated disc, ICD) and some channels, such as connexins, are specifically targeted to this region (749). Some channels, such as the cardiac Na+ channel subunit Nav1.5, are targeted both to lateral membranes and to the ICD and exist as two distinct pools of channels that interact with specific scaffolding proteins (735). The lateral Na+ channel exists as a complex with syntrophin and dystrophin, whereas the pool of Na+ channels in the ICD associates with desmosomal proteins (735). A similar “two-pool” model is possible for KATP channels as suggested by the finding that dystrophin is required for the cardioprotective function of KATP channels in cardiomyocytes (280). Moreover, there is emerging evidence that KATP channels are expressed at a higher density at the ICD (356). The KATP channels at the ICD do not appear to have distinct electrophysiological properties, but they interact and colocalize with the desmosomal proteins plakophilin-2 and junction plakoglobin. The functional roles for KATP channels at the ICD are unclear, but they may be involved in cardiac electrical conduction (356) and with arrhythmias associated with arrhythmogenic cardiomyopathy, which results from mutations in desmosomal proteins (176).
KATP channels are fairly stable proteins. Pulse-chase experiments have demonstrated that SUR1 is stable when expressed alone, but that Kir6.2 is rapidly degraded with a half time of ∼30 min. When Kir6.2 is expressed with SUR1, the channel complex has a half-life of ∼8 h (138). Another study has determined that the half-life of Kir6.2/SUR1 channels is ∼15 h, whereas the Kir6.2/SUR2A might be more stable with a half-life of ∼20 h (47), despite their colocalization with LAMP2, a marker of late endosomes and lysosomes (see sect. VF). There is some evidence that under certain conditions KATP channels can be diverted to a lysosomal degradation pathway since lysosomal inhibitors prevents protein kinase C (PKC)-induced degradation (538). It is likely, however, that constitutive degradation occurs via the proteasome since nonstimulated degradation is unaffected by the lysosomal inhibitor chloroquine and prevented by MG132, which is an inhibitor of the proteasome (788, 858, 891).
I. The Physiological and Pathophysiological Relevance of KATP Channel Trafficking
A clear role for leptin receptor signaling and glucose in KATP channel trafficking in the pancreatic β-cell has recently emerged. (109). Low extracellular glucose levels, associated with fasting, increase the KATP channel surface density in an AMPK-dependent pathway (504). Leptin, a hormone secreted by adipose cells, acts centrally in the hypothalamus to control food intake. It has additional effects on the pancreatic β-cell to reduce insulin secretion. Recent data indicate that this action of leptin is attributed to stimulating KATP channel to the surface membrane and that this pathway is controlled by AMPK, PKA, and Ca2+/calmodulin-dependent protein kinase kinase β (109, 632).
The role of KATP channel subcellular trafficking has not been studied extensively in the cardiovascular system, and conflicting data exist. In guinea pig cardiomyocytes, hypoxia/reoxygenation before a sustained hypoxic episode (a cellular model of ischemic preconditioning) was shown to improve cellular survival and also to increase the amount of Kir6.2 surface protein and the KATP channel current density (93). Pharmacological preconditioning with epinephrine had a similar effect in isolated rat ventricular myocytes (824). The increased KATP channel density (presumably due to enhanced trafficking) was mediated by AMPK, PKCδ, PKCε, and p38 MAPK (767, 824). PKC was also implicated in the elevated levels of KATP channel subunits in sarcolemmal fractions after ischemia/reperfusion in the female rat heart (199). The findings that PKC increases KATP channel density in hearts is somewhat perplexing since most cellular studies suggest that PKC causes KATP channel internalization or degradation (36, 363, 538). Ca2+/calmodulin kinase II (CaMKII) may also have a role in KATP channel endocytic recycling, particularly during cardiac ischemia. CaMKII inhibition was shown to increase the KATP channel surface density, which was linked to improved ventricular myocyte survival during ischemia (495). A subsequent study demonstrated that CaMKII interacts with and phosphorylates the Kir6.2 subunit (at Thr-180 and Thr-224) to promote endocytosis (449, 741). Although some studies suggested that KATP channel surface density is increased with cardiac ischemia (47, 199), it is tempting to speculate that prolonged or severe ischemia that is associated with an increased cytosolic Ca2+ may be associated with KATP channel endocytosis, mediated by a Ca2+-induced activation of CaMKII. This possibility remains untested.
VI. THE PHARMACOLOGY OF KATP CHANNELS
There are many inorganic compounds that can promote or inhibit KATP channel opening. These are respectively referred to as KATP channel openers (KCO) or inhibitors/blockers. The pharmacological properties and a detailed description of the structures of these compounds are beyond the scope of this text, and the reader is referred to reviews for more information in these topics (200, 223, 645). The chemical structures of some of the compounds are shown in Figure 9. Here, we will list examples of some of these compounds commonly used in the literature (Table 3) and review some of the mechanism(s) by which they affect KATP channel opening.
A. KATP Channel Inhibitors
An ion such as Ba2+ blocks the KATP channel by restricting movement of K+ through the channel′s permeation pathway (779) and can therefore be referred to as a channel blocker. The block of KATP channels by TEA and 4-aminopyridine (4-AP) most likely share the same mechanism (416). Strictly speaking, compounds that inhibit KATP channel function through other mechanisms should be referred to as KATP channel inhibitors (or antagonists), but we will interchangeably refer to them as inhibitors and blockers. The most commonly used class of KATP channel inhibitors (for the treatment of type II diabetes) is the hypoglycemic sulfonylureas, which includes tolbutamide and glibenclamide. One of the first studies to suggest that tolbutamide may affect K+ fluxes in the pancreatic β-cell was made in the late 1970s with 86Rb+ flux assays (343). Soon after its discovery in pancreatic β-cells (135), the KATP channel was identified as the receptor for sulfonylureas (193, 766). All types of KATP channels are now known to be inhibited by sulfonylureas (e.g., tolbutamide and glibenclamide). The sensitivities of diverse KATP channels to sulfonylureas differ; for example, the KATP channels in the endothelium, vascular smooth muscle, and the cardiac conduction system are more efficiently blocked by tolbutamide than the KATP channel in the atrium or ventricle (Table 3). Other compounds also inhibit KATP channel function, including certain saturated fatty acids, such as the decanoic acid, 5-hydroxydecanoic acid (5-HD), which blocks the “ventricular” Kir6.2/SUR2A channel with an IC50 of 0.2–30 μM in inside-out patches (497, 609). 5-HD also blocks KATP channels in mitochondria (262) with an IC50 value of 45–75 μM (386). 5-HD may have complex actions, since the compound blocks ventricular KATP channels in isolated membrane patches, but is less effective in intact cells (497). 5-HD is readily activated to 5-HD CoA, which is taken up into the mitochondrial matrix and is metabolized by the β-oxidation pathway, albeit with slow kinetics (320–322). 5-HD also inhibits β-oxidation of other fatty acids (320), which will affect cardiac energy metabolism. Other compounds can also inhibit KATP channel function. For example, HMR 1098 is often used as a selective cardiac sarcolemmal KATP channel blocker (199, 512, 539), but a recent systematic study demonstrated that the compound affects various classes of KATP channels (921). Care should therefore be employed when using any of these compounds experimentally.
1. Where do sulfonylureas bind?
Sulfonylureas bind to SUR1 with high affinity. The Kd for glibenclamide binding is ∼0.5–10 nM (4, 186). In contrast, the sulfonylurea binding affinity for SUR2x is somewhat lower. The Kd for glibenclamide binding to SUR2A and SUR2B is similar at ∼30–300 nM (186, 674, 675). When coexpressed with Kir6.1, however, the SUR2B binding affinity increases from 32 to 6 nM (674). Higher sulfonylurea concentrations (3- to 7-fold) are needed to inhibit channel function (186, 674), and simultaneous occupancy of four binding sites may be needed for channel inhibition (674). The distal 42 amino acids, which is the only difference between these two SUR2x splice variants, are not involved in binding, since a chimeric SUR2 construct in which this COOH-terminal region was replaced with that of SUR1 did not have a higher sulfonylurea binding affinity (186). The high-affinity sulfonylurea binding site was identified by generating chimeras between the high (SUR1) and low (SUR2A) affinity receptors. SUR2A became highly sensitive to tolbutamide inhibition when replacing its transmembrane domains 14–16 with that from SUR1 (26). Mutating a single residue in this region of SUR1 TMD2 (S1237) was sufficient to prevent high-affinity tolbutamide inhibition. Conversely, exchanging the corresponding residue in SUR2B for a serine increases the affinity of glibenclamide binding 5–10 times (314, 518). The sulfonylurea binding pocket involves other regions of the protein as well. Deletion analysis has defined regions dispersed throughout SUR that influence glibenclamide binding, including the cytosolic loop betweenTMD0 and TMD1, and the binding pocket is mostly likely only revealed in the correctly folded hetero-octameric KATP channel (563).
2. Modulation of sulfonylurea inhibition by intracellular nucleotides
Intracellular nucleotides modulate the degree of KATP channel inhibition by sulfonylureas. The effect of MgADP is most striking, and interestingly, the effect on different types of KATP channels is opposite in nature. In the pancreatic β-cell, MgADP increases the efficiency by which tolbutamide inhibits the KATP channel by 10-fold (286, 707, 947). Chimeric constructs made from SUR1/SUR2 identified a region of SUR1 that includes transmembrane (TM) regions 8–11 as being important in the ability of MgADP to potentiate sulfonylurea inhibition (659). One possibility is that sulfonylureas impair the ability of MgADP to stimulate Kir6.2/SUR1 channel activity, and consequently that the inhibitory effects of ADP on Kir6.2 now predominate (286). This may happen either by displacing MgADP from its binding site, or by preventing intramolecular transduction of the stimulatory effect of MgADP (659). In SUR2-containing KATP channels, however, MgADP impairs the ability of sulfonylureas to inhibit KATP channel function (313, 711, 842). For example, the SUR2A-based cardiac KATP channel is half-maximally inhibited by 0.5 μM glibenclamide when recordings are made in the absence of ADP. However, in the presence of 100 μM MgADP, glibenclamide is ineffective, even at concentrations as high as 300 μM glibenclamide (842). Moreover, glibenclamide inhibition of whole-cell KATP channel current is more effective when ADP is absent from the pipette solution (842). These data might explain the finding that that sulfonylurea block of cardiac KATP channels is impaired (or even prevented) under conditions of metabolic inhibition (228, 842).
B. KATP Channel Openers
Examples of KCOs include nicorandil, minoxidil, pinacidil, cromakalim, and diazoxide. Several classes of KCOs exist (200), and electrophysiologically, they can be classified into distinct groups (801). Some KCOs, which includes pinacidil, cromakalim, and derivatives, decrease the sensitivity of the KATP channel to ATP, resulting in increased channel opening at a given level of cytosolic ATP. A compound in this group, pinacidil, does not activate SUR1-containing KATP channels (such as in the pancreatic β-cell), but shows little discrimination amongs cardiovascular KATP channel subtypes (Table 3). KCOs in another group, including diazoxide and nicorandil, require the presence of intracellular ADP for activity. Nicorandil, which is a nicotinamide nitrate derivative, has a complex pharmacological profile that combines nitrate-like and KATP channel opening activities, and the results obtained with this compound are not always easy to interpret mechanistically. Another compound in this group, the benzothiadiazine diazoxide has been studied extensively in the cardiovascular literature. It was originally described as having vasodilatory activity (671), but it soon became apparent that it also causes hyperglycemia (616, 878) by inhibiting insulin release from pancreatic β-cells (344). We now know that these effects are due to KATP channel activation in these tissues (755, 816). In fact, several subtypes of KATP channels have a high sensitivity to diazoxide (K1/2 in the micromolar range; Table 3) (535, 594, 816, 946). KATP channels in some tissues, such as the ventricle, are relatively insensitive to diazoxide under basal conditions, but the channel is sensitized to the compound under metabolically impaired conditions (151, 546).
1. Where do KATP channel openers bind?
KCOs act on the SURx subunit as demonstrated by coexpression studies in Xenopus oocytes, where Kir6.1 currents were shown to be activated by diazoxide only when coexpressed with SUR1 (15). The SURx isoform present correlates with the channel′s pharmacological profile. For example, channels containing SUR1 are strongly activated by diazoxide, but not by cromakalim or pinacidil (283, 284, 598). In contrast, SUR2A-containing channels respond weakly to diazoxide under basal conditions, but are potently activated by pinacidil (284, 573, 731). Channels containing the SUR2B splice variant are sensitive both to diazoxide and cromakalim (711). Chimeric constructs between SUR1 and SUR2A identified transmembrane domains (TMDs) 6–11 and the first nucleotide-binding domain (NBD1) of SUR1 to be necessary for channel activation by diazoxide (39) (Figure 10).
2. Modulation of KCO activity by intracellular nucleotides
Mutagenesis in either NBF1 (285) or NBF2 (739) impairs the ability of diazoxide to activate Kir6.2/SUR1 channels. Moreover, diazoxide stimulation of pancreatic β-cell or Kir6.2/SUR1 channel activity was found to require the presence of Mg2+ and hydrolyzable nucleotides, MgATP or MgADP (479, 739). It has been suggested that the ability of diazoxide to activate KATP channels is dependent on the ATPase activity of the SURx NBF domains (711) and that diazoxide may potentially increase KATP channel activity by stimulating the ATPase activity of the channel complex (285). The difference in diazoxide sensitivity between SUR2A and SUR2B is interesting and informative, since these variants only differ from each other in the distal COOH-terminal tail (42 amino acids). In an elegant study, it was indeed found that the distal COOH-terminal tail is necessary for channel activation (Figure 10), both by MgADP and diazoxide (546). Each of the Kir6.2/SUR1, Kir6.2/SUR2A, and Kir6.2/SUR2B channels was activated by diazoxide, but the activation of Kir6.2/SUR2A was only observed at elevated ADP levels. The importance of the COOH terminus was demonstrated when swapping the distal COOH terminus of SUR2A with that of SUR1, which formed a KATP channel that responded effectively to MgADP and diazoxide (546). These data suggest that the SUR2A COOH terminus might interfere with MgADP binding to NBF2, which is why higher cytosolic ADP concentrations are needed before diazoxide will activate the SUR2A-containing channel. Residues within NBF2 are also important, as demonstrated by a recent study by the Light group (219). They found that mutagenesis of a single amino acid residue in the SUR1 NBF2 (S1369) to the corresponding positively changed residue of SUR2 (K1337) is sufficient to decrease the ATPase activity and to prevent diazoxide-induced activation of the mutant SUR1 channel in the absence of ADP (219).
C. Issues With the Use of KATP Channel Modulating Compounds
We have seen that the efficiency of commonly used KATP channel openers and inhibitors depend on the levels of intracellular nucleotides, with MgADP playing a particularly important role. Thus a potential for problems exists when using these compounds to assess the role(s) of KATP channels in a setting of a changed metabolic status. This would be particularly problematic during hypoxia and ischemia when cellular MgADP levels are increased. For example, increasing the cytosolic MgADP concentration renders glibenclamide ineffective as a KATP channel blocker (228, 842). Similar problems exist with some of the KATP channel openers. Although diazoxide has almost no effect on the “ventricular” Kir6.2/SUR2A channels in excised patches in the absence of MgADP (371), it is known to shorten the cardiac action potential (118, 580, 694) (see sect. IXB2). As stated in the preceding paragraph, one reason for these discrepant observations is that diazoxide only becomes active when MgADP is elevated (SUR2A/Kir6.2 channels become as sensitive to diazoxide as SUR1/Kir6. 2 channels when 100 μM MgADP is present in the “cytosolic” solution; Ref. 151). Thus care should be taken when using diazoxide as a compound with specificity or selectivity for subtype(s) of KATP channels. These issues have been discussed in a recent review (125).
Not only does glibenclamide become an inefficient KATP channel blocker in metabolically impaired cells (230) and diazoxide a more efficient opener (151), but each of the KATP channel modulators also have unanticipated off-target effects (Table 3). For example, tolbutamide uncouples oxidative phosphorylation of mitochondria (428). It may also protect the ischemic myocardium by increasing glycolysis without increasing lactate production by promoting the entry of pyruvate into the mitochondria (461), which results in an ATP-sparing effect (476). Glibenclamide blocks other types of channels, and most of the KATP channel modulators directly or indirectly affect mitochondrial function (Table 3), including uncoupling of oxidative phosphorylation, blocking succinate dehydrogenase, and/or serving as substrates for mitochondrial respiration. These issues are discussed in more detail later (see sect. IXB7).
VII. DIFFERENT SUBTYPES OF KATP CHANNELS IN THE CARDIOVASCULAR SYSTEM
A. Ventricular KATP Channels
1. Biophysical properties of native channels
The conductance of the single KATP channel exhibits weak inward rectification (inward currents are correspondingly larger than outward currents). In general, the slope conductance is determined for inward currents at a linear region of the current-voltage (I-V) relation. Slope conductance values of the ventricular KATP channel in symmetrical high (140 mM) K+ concentrations have been reported to be 75–80 pS (38, 48, 317, 416). In the absence of nucleotides, the KATP channel opens with a high (∼0.82) Po (232). Opening occurs with a bursting behavior (359), with intraburst open and closed times, respectively of ∼1.8 ms and 0.15 ms at −80 mV (38). The dwell-time kinetics are dependent on the membrane voltage and to some extent also the K+ driving force (938).
The Po of the ventricular KATP channel is strongly modulated by the presence of nucleotides at the cytosolic face of a membrane patch (416, 607, 815). The ATP concentration needed to produce half-maximal block (IC50) varies between studies, and values between 20 and ∼100 μM have been reported (Table 2). There is a large variation in reported IC50 values, even from the same laboratory; for example, the IC50 values for different patches from rat ventricular myocytes exhibit a 60-fold difference, ranging between 9 and 580 μM (232). Moreover, the channel′s ATP sensitivity is not fixed. Even in the same patch, the IC50 value can change from 215 to 46 μM within minutes after patch excision (127). Although the KATP channel can be blocked by high cytosolic ADP levels, it is stimulated by MgADP, and the concept has evolved that the channel is regulated by the ATP/ADP ratio. For a more complete discussion of the nucleotide regulation of KATP channels, please refer to section II.
2. Molecular composition
The current consensus is that the ventricular KATP channel protein complex minimally consists of a Kir6.2/SUR2A subunit combination. An argument in favor of this notion includes the expression of Kir6.2 and SUR2A mRNA and protein in ventricular tissue (117, 372, 579, 684). Knockdown of Kir6.2 and SUR2 expression by antisense oligonucleutides decreases expression of native ventricular KATP channels (904). Likewise, overexpression of dominant-negative Kir6.x subunits impairs ventricular KATP channel function (474, 810, 830, 936), and ventricular KATP channels are absent in knockout mice genetically deficient of Kir6.2 or SUR2 subunits (763, 769). Heterologously expressed Kir6.2/SUR2A channels have properties similar to native ventricular KATP channels, such as the single-channel conductance, dwell-time kinetics, sensitivity to block by intracellular ATP, and stimulation by intracellular MgADP (38). Moreover, heterologously expressed Kir6.2/SUR2A channels have a pharmacological profile similar to native ventricular KATP channels, including their block by glibenclamide and stimulation by cromakalim and pinacidil (in the presence of intracellular MgATP) and relative insensitivity to diazoxide when intracellular MgADP is absent (38, 151, 546). There are reports that Kir6.1 (525, 540, 565, 578, 881) and SUR1 (578) mRNA and protein are expressed in ventricular myocytes. Although genetic knockout of Kir6.1 does not appear to affect the whole cell ventricular KATP channel current (565), the roles of Kir6.1 in ventricular function are not understood. There is a possibility that Kir6.1 may contribute to a novel small (21 pS) conductance KATP channel in the ventricle, which has similar properties to the 80 pS channel, but has a high sensitivity to diazoxide (881). Thus the contribution of the “atypical” KATP channel subunits cannot be completely discounted, especially considering reports demonstrating upregulation of Kir6.1 expression levels in the ventricle after treatment with KATP channel opening drugs or after cardiac ischemia (74, 381). Please refer to section VA.
3. Physiological roles of ventricular KATP cardiac channels
In patch-clamp experiments, there is little evidence for basal or spontaneous opening of KATP channels in resting ventricular myocytes. Although some reports demonstrate that KATP channel block with glibenclamide or tolbutamide lengthens the ventricular action potential duration (118, 213), suggestive of constitutively active KATP channels, most reports show that the ventricular action potential is relatively insensitive to glibenclamide or 5-HD under basal conditions (133, 400, 577, 747). Blockade of KATP channels with glibenclamide or HMR-1098, however, attenuates the action potential duration shortening induced by KATP channel openers, metabolic inhibition, hypoxia, and ischemia in a variety of animal models and experimental preparations (Table 6), which gave rise to the concept that ventricular KATP channels only open under conditions of metabolic impairment. For a full discussion of this topic, please refer to section IXB2. Recent data with genetic mouse models have demonstrated that ventricular KATP channels do have important physiological roles, including the provision of tolerance to exercise and stress. They also contribute to the action potential duration adaptation that occurs over a time course of a few minutes upon changes in heart rate (see sect. VIII).
B. Atrial KATP Channels
1. Biophysical and pharmacological properties of native atrial KATP channels
Soon after the identification of the ventricular KATP channel, KATP channels were also identified in the atria of guinea pig (310) and human hearts (342). These channels are superficially similar to ventricular KATP channels with a unitary conductance between 52 and 80 pS (Table 2). They exhibit bursting behavior with intraburst mean open and closed times of 1.4 and 0.3–0.6 ms, respectively (50, 342). Atrial KATP channels are activated by hypotonic solutions or by membrane stretch (50, 833). Atrial KATP channels have a distinct pharmacology. Although the rabbit and guinea pig atrial KATP channel is activated by 10–500 μM nicorandil (310, 485), rilmakalim (7 μM) and pinacidil (100 μM) were described not to activate the atrial KATP channel in the rabbit atrium (668). Pinacidil was found to block the atrial KATP channel in other studies (Table 3). In atrial appendages from rat heart, KATP channels are blocked by glibenclamide with an IC50 value of 1.2 nM (50, 310) and activated by low diazoxide concentrations (50, 237); these are similar potencies as the high-sensitivity glibenclamide block or diazoxide activation of pancreatic β-cell KATP channels (460, 703). Tolbutamide blocks rodent atrial KATP channels with high affinity (833), but the human atria KATP channel is relatively insensitive to tolbutamide (945).
2. Molecular composition of atrial KATP channels
Neonatal rat atrial appendage cardiomyocytes were reported to express mRNA for Kir6.1, Kir6.2, SUR1A, SUR1B, SUR2A, and SUR2B subunits (as determined by RT-PCR) (50). A role for Kir6.x subunits is demonstrated by the finding that overexpression of dominant-negative Kir6 subunits suppress native KATP channels in rat atrium (830). Moreover, KATP channels are absent in the atria isolated from Kir6.2−/− mice (678). A side-by-side comparison of mouse tissues revealed that SUR1 protein is predominantly expressed in the atrium (and to a lesser extent in the ventricle) (237). Moreover, the KATP channel current is virtually absent in atria from SUR−/− mice, whereas the ventricular KATP channel is unaffected (237). The SUR1 composition of atrial KATP channels appears therefore to be responsible for the unusually high diazoxide sensitivity in this tissue (Table 3). It should be noted that this situation may be unique to rodents since the differential sensitivity to diazoxide in the atria and ventricles of dog and human heart is less prominent (222, 922).
3. Physiological roles of atrial KATP cardiac channels
Atrial KATP channels most likely have similar roles to those in the ventricular myocardium, namely, protection against stress responses and action potential duration adaptation to sudden increases in the heart rate. If anything, the roles of KATP channels in the atrium are likely to be more pronounced since atrial KATP channels are significantly more sensitive to metabolic inhibition in atrial than ventricular myocytes (642). An additional role for KATP channels has been suggested in the release of atrial natriuretic peptide (ANP) from atria. However, pharmacological data are inconsistent. Some studies demonstrated that KATP channel block with tolbutamide and glibenclamide suppress the release of atrial natriuretic peptide (ANP) from rat and dog atria in response to mechanical stretch (437) or heart failure (107). Other studies, however, suggested that 100 μM tolbutamide increases the release of ANP from neonatal rat atrial myocytes or intact atria in response to stretch or hypoxia (402, 886). Moreover, the stretch-induced ANP secretion is inhibited by KATP channel openers pinacidil or diazoxide (886). The mechanosensitive modulation of the atrial KATP channel (833) by membrane stretch may well be implicated in this process. Definitive evidence for a role of KATP channels in ANP release from atria was obtained with Kir6.2−/− mice, in which volume expansion has a more pronounced effect on the elevation of plasma ANP concentration and induction of hypotension compared with wild-type mice (678). Collectively, these data are consistent with a role for mechanical stretch to open KATP channels, which are linked to ANP secretion from the atria, which in turn affects a variety of physiological end points, including vasodilation, anti-ischemic actions, and reduced hypertrophic effects.
C. KATP Channels in the Specialized Cardiac Conduction System
1. Biophysical and pharmacological properties of native channels
The literature hints at the possibility that cardiac conduction system (CCS) KATP channels may have a unique molecular composition (which determines function and pharmacological properties). For example, although the AV nodal KATP channel is superficially similar to the ventricular channel, it was reported to have a smaller conductance (415). Similarly, myocytes from the rabbit sinoatrial (SA) nodal and Purkinje myocytes also have a smaller unitary conductance (52–60 pS; cf. ∼70–80 pS in the ventricle; Table 2). Moreover, the ATP sensitivity of rabbit Purkinje KATP channels is lower than that of the ventricular channel (48, 502). Using a novel reporter mouse that expresses EGFP specifically in the CCS (628), we examined the properties of CCS KATP channels and found significant and biologically relevant differences between CCS and ventricular KATP channels (48). Aside from a smaller unitary conductance, we observed differences in their nucleotide regulation. Not only were the CCS KATP channels less sensitive to inhibitory ATP, but they were more readily activated by intracellular MgADP. Ventricular KATP channels were characteristically stimulated by lower MgADP concentrations (227), but inhibited as the ADP concentration was increased (416). In contrast, the CCS KATP channels were stimulated by MgADP, but not inhibited within the concentration range examined. Half-maximal activation occurred at 84 ± 3.7 μM MgADP (48). These results suggest that CCS KATP channels are preferentially opened by alterations in intracellular nucleotides, which may be a pathophysiologically relevant finding given the importance of ventricular CCS and Purkinje-ventricular junction (PVJ) in the generation of ischemia-induced arrhythmias (392) (see sect. IXB4 for further discussion of this topic).
Glibenclamide has no effect on the basal heart rate of isolated rat hearts (424) or the firing rate of SA nodal cells isolated from rabbit hearts (318), suggesting that KATP channels are constitutively closed. KATP channel openers such as nicorandil and cromakalim, in contrast, decreases the cardiac pacemaker rate, shortens the SA nodal action potential and hyperpolarizes the membrane potential, and may produce atrioventricular (AV) block (278, 697). In inside-out membrane patches excised from rabbit SA node pacemaker cells, cromakalim or pinacidil activates and glibenclamide blocks KATP channels (318). Levcromakalim also shortens the action potential duration of rabbit Purkinje fibers (502). We demonstrated that diazoxide (200 μM) is more effective to activate mouse CCS KATP channels than ventricular KATP channels, whereas levcromakalim had the opposite profile (48). Thus CCS KATP channels may have a different functional and pharmacological properties compared with the ventricular KATP channel.
2. Molecular composition
A large-scale transcriptional analysis of mouse cardiac tissue showed that Kir6.1, Kir6.2, and SUR2 mRNAs were expressed at higher levels in the ventricle compared with the SA node or AV nodes (540). In the dog and mouse, we found elevated expression of Kir6.1 mRNA and protein (and decreased expression of Kir6.2) in the CCS relative to the ventricle (48). A role for Kir6.1 in cardiac conduction is underscored by the finding AV block is observed in radio telemetry recordings of the ECG in Kir6.1−/− mice (565). Although the role of Kir6.1 and the mechanism(s) by which Kir6.1 subunits may contribute to cardiac conduction are presently unclear, clinical data demonstrate that a strong association exists between genetic variation in KCNJ8 and the risk of Brugada syndrome, J wave syndromes, ventricular fibrillation, and atrial fibrillation (see sect. X). An important role also exists for Kir6.2, since the Kir6.2−/− mice lack SA nodal pinacidil-activated KATP channel currents, and hypoxia-induced bradycardia is mitigated in the knockout mice (248). We found the contactin 2-positive CCS mouse myocytes to be enriched for SUR2B, whereas SUR1 was undetectable (48). The elevated SUR2B expression might account in part for the enhanced sensitivity of the mouse CCS to diazoxide (546). The possibility of diversity of KATP channel subunit expression in various regions of the CCS (e.g., proximal vs. distal) is real, but has not been formally investigated.
3. Physiological and pathophysiological roles of KATP channels in the specialized cardiac conduction system
KATP channels appear to be expressed in all regions of the cardiac conduction system, including the SA node, AV node, and the specialized Purkinje network. Their activity affects the functional properties of these cells (415, 502, 696). For example, hypoxia produces bradycardia, which is mediated by KATP channel opening (248, 498). The conduction slowing and AV block that occurs with hypoxia in Langendorff-perfused rabbit hearts is also mediated by KATP channels since it can be prevented by glibenclamide (698). Glibenclamide has also been described to reduce the intramyocardial conduction delay induced by ischemia in dog (61) and mouse (48) heart, and to decrease ischemia-induced transmural conduction block in a canine wedge preparation (633). Similarly, glibenclamide prevents the increased longitudinal resistance (electrical uncoupling) in an isolated rabbit septal preparation during ischemia (787) and prevents the subendocardial conduction delay produced by ischemia in an isolated, arterially perfused canine interventricular septal preparation (576). Thus CCS KATP channels are likely to participate in ischemia-induced conduction disturbances and precipitate life-threatening arrhythmias, and they may therefore represent novel therapeutic targets.
D. KATP Channels in the Vascular Smooth Muscle
1. Biophysical properties of native channels
Vasodilators such as diazoxide (671) and pinacidil (21) hyperpolarize smooth muscle cells due to an increased K+ conductance (345). The initial finding that pinacidil activated KATP channels in the heart (19) suggested the possibility that KATP channels were also present in vascular smooth muscle. Indeed, patch-clamp recordings demonstrated that single KATP channels are present in excised patches from smooth muscle cells enzymatically dissociated from rabbit and rat mesenteric arteries (755). Single KATP channels had a fairly low open probability, were inhibited by cytosolic ATP in the submillimolar range, and had a unitary conductance of ∼135 pS. The channels were activated by pinacidil, diazoxide, and cromakalim and blocked by glibenclamide (755). In subsequent studies, two types of vascular KATP channels have been described: either with a small/medium (10–50 pS) or large (135–200 pS) conductance (the reader is referred to an excellent review on this subject, see Ref. 649). The small/medium conductance channels require the cytosolic presence of nucleoside diphosphates (NDP) for opening and are therefore often referred to as KNDP channels. The vascular KNDP channels can be activated by metabolic inhibition (75, 919); cellular ATP depletion; hypoxia (154); KATP channel openers such as levcromakalim (75), pinacidil (274, 413), and nicorandil (377, 414, 853); and inhibited by KATP channel blockers such as glibenclamide (75, 853, 919). The presence of Ca2+ appears to be required for KNDP channel opening (377, 853). Vascular KNDP channels open in bursts with at least two open states and several closed states (414, 853).
2. Molecular composition
Transcripts of the KATP channel subunits Kir6.1 and SUR2B are highly expressed in vascular tissue, whereas SUR1 and SUR2A mRNAs are expressed at lower levels (117, 372, 374, 382). Kir6.2 mRNA is present in several types of smooth muscle (277, 378, 395, 662, 743, 799). However, despite the presence of the transcript, the Kir6.2 protein is not expressed [e.g., in human coronary artery (907)] and the possibility of posttranscriptional regulation of Kir6.2 expression must be considered. Moreover, genetic deletion of Kir6.2 in mice abolishes KATP channels in the cardiac ventricle, but not in the aorta, further demonstrating that this subunit has little role in the vascular smooth muscle of the mouse (769). Heterologous coexpression of Kir6.1 and SUR2B subunits gives rise to channels with functional and pharmacological properties that recapitulate smooth muscle KNDP channels [e.g., a single-channel conductance of ∼33 pS, a requirement for intracellular NDPs for channel opening, and a relative insensitivity to block by ATP (888)]. Moreover, both Kir6.1−/− and SUR2−/− mice lack KATP channel activity in aortic smooth muscle myocytes (115, 565), lending further support to the idea that smooth muscle KATP channels are composed of a Kir6.1/SUR2B subunit combination. Both mouse models additionally exhibit incidences of spontaneous coronary vasospasm, resembling human Prinzmetal angina. Interestingly, however, this phenotype is not eliminated when reintroducing SUR2B specifically into smooth muscle in the SUR2−/− background, suggesting that the coronary vasospasm may arise from a smooth muscle-extrinsic process (417).
3. Physiological roles of vascular KATP channels
The vasculature, particularly the small resistance arterioles, is under constant tone to regulate blood flow to the periphery and to organs. The vascular tone is not constant and is subject to moment-to-moment variations to regulate blood as needed. In some tissues, such as the heart and the brain, spontaneous vasodilation and vasoconstriction may occur upon changes in metabolic demand (named autoregulation of blood flow). Vascular tone and blood flow may also be regulated by a number of vasoactive substances such as calcitonin gene-related peptide, adenosine or prostacyclin (which cause vasodilation) or angiotensin II, vasopressin, or endothelin (which cause vasoconstriction). Opening of KNDP channels hyperpolarizes smooth muscle cells, which leads to less Ca2+ entry, vasodilation, and increased blood flow (649) (Figure 11). The contribution of KNDP channels to regulating the membrane potential of smooth muscle cells (and thus blood flow) increases as the coronary diameter decreases along the vascular bed (7, 106, 198, 447, 587, 691), consistent with a significant role for KATP channels in the smaller resistance vessels (686). Their contribution to the regulation of coronary function was first identified by the vasodilatory actions of KATP channel openers (290, 649, 682) and the vasoconstrictive effects of KATP channel blockers (424). Thus an essential role for coronary KATP channels was identified in the vasodilatory responses of adenosine (155, 649), angiotensin II (64), hypoxia (161, 649), exercise (192), and metabolic demand (33, 34, 427), and it is now accepted that KATP channels are essential for maintaining basal coronary vascular tone in vivo (454, 686).
E. KATP Channels in the Endothelium
1. Biophysical and pharmacological properties of endothelial KATP channels
KATP channels are not only present in the coronary smooth muscle cells, but also in the coronary endothelium (380, 469, 508, 708, 851, 946). The endothelial KATP channels differ from the vascular channels in terms of their molecular composition, pharmacological profiles, biophysical properties, and functional roles. Both cells types express many different ion channels (221, 388, 601) with diverse roles in the regulation of blood. In resistance vessels, the endothelium may regulate smooth muscle contraction (and thus blood flow and blood pressure) by 1) direct effects on the smooth muscle membrane potential (Em) through electric coupling via gap junctions and/or 2) by releasing of vasoactive substances in response to various stimuli. Ion channels (including KATP channels) are involved in both of these processes. Endothelial cells have fairly negative resting Em of between −80 and −30 mV, depending on their origin and treatment (culturing vs. freshly isolated, etc.) (2). There are reports of diversity within a single cell population. For example, bovine aortic endothelial cells have a bimodal distribution, with some cells having a resting potential around −25 mV whereas others are at −85 mV (558). Interestingly, transitions between these two states can occur due to the N-shaped I-V relationship of endothelial cells. The relationship between the Em and extracellular K+ is well described by the Nernst equation (52 mV/decade change in [K+]o), with some deviation at lower K+ concentrations, suggesting a strong basal K+ conductance (2). Indeed, most published studies report the presence of inward rectifier K+ channels (2, 602).
The description of endothelial KATP channel properties and molecular composition is scant, but convincing. The majority of reports demonstrate that KATP channel openers (levcromakalim, rimakalim, pinacidil, and minoxidil) cause membrane hyperpolarization of various magnitudes in endothelial cells from rabbit aorta, guinea pig isolated capillary fragments and human coronary artery (426, 477, 478, 907; but see Ref. 558). In whole cell patch-clamp experiments performed with freshly isolated cerebral microvascular and rat aortic endothelial cells, a glibenclamide-sensitive K+ current is activated by intracellular dialysis with low [ATP], with metabolic inhibition or by pinacidil (390). Essentially similar results were found with freshly isolated rabbit aortic endothelial cells (426) and isolated guinea pig capillaries (705). Basal KATP channel activity is suggested by the finding that in bovine pulmonary endothelial cells glibenclamide inhibits a proportion of the current (in the absence of KATP channel agonists or metabolic inhibition) (105). Interestingly, the glibenclamide-sensitive component of the IV relationship is hugely increased when endothelial cells are adapted to flow conditions (likely to better represent a physiological setting compared with cells grown under static culturing conditions) (105). Effects of KATP channel blockers on the Em have not been investigated. At the single-channel level, one report suggests the presence of a 40 pS KATP channel, which is blocked by cytosolic ATP and glibenclamide, in isolated patches of cerebral microvascular endothelial cells (390). Another reports the presence of two types of channels (25 and 150 pS), both being blocked by cytosolic ATP or glibenclamide (426). Endothelial KATP channels appear to be exquisitely sensitive to diazoxide and tolbutamide (426, 477, 478) (Table 3).
2. Molecular composition
The molecular composition of endothelial KATP channels remains to be fully characterized. Guinea pig cardiac capillaries express Kir6.1, Kir6.2, and SUR2B mRNA (705). These subunits are also present in human coronary artery endothelial cells, both at the mRNA and protein level (907). In rats, the endothelial KATP channel was described to be of a SUR2B/Kir6.1 subtype (793). Pulmonary artery endothelial cells isolated from Kir6.2−/− mice have an impaired response in membrane potential changes and ROS generation associated with shear stress (547). Moreover, transgenic expression of Kir6 dominant negative subunits specifically in the endothelium lead to impaired coronary artery function, elevated blood pressure, and defects in ET-1 release (536). SUR1 does not appear to be expressed at significant levels under physiological conditions, but there is evidence of enhanced endothelial SUR1 expression following stroke, where it may participate in cerebral edema (744).
3. The physiological role of endothelial KATP channels
Prevailing evidence suggests that KATP channels in the coronary microvasculature are responsible for maintaining a constant blood flow to the myocardial tissue (387, 388). They regulate coronary flow alterations in response to metabolic demand (160) and the maintenance of flow in response to changes in blood pressure (454, 589). There is evidence that endothelial KATP channel opening may improve pressure overload-induced cardiac remodeling in rats by protecting endothelial function (793). Their contribution increases as the coronary diameter decreases along the vascular bed, consistent with a significant role for KATP channels in the smaller resistance vessels (7, 447, 587, 686, 691). In the vascular smooth muscle cells, KNDP channels regulate the membrane potential and the contractile state of the smooth muscle myocytes, but endothelial KATP channels add another dimension to the regulation of blood flow (Figure 11). For example, endothelial KATP channels mediate coronary microvascular dilation to hyperosmolarity (380). An important link to adenosine and NO generation also appears to exist. For example, in the isolated perfused guinea pig heart endothelial KATP channels have been implicated in the response to adenosine and extracellular ATP (488). With the use of cannulated microvessels isolated from porcine heart, it was confirmed that the vasodilatory effect of adenosine (particularly at lower doses) is mediated by endothelial KATP channels (469). Adenosine-induced vasodilation is also attenuated by l-arginine analogs that inhibit NO synthesis (488) or by the eNOS inhibitor NG-monomethyl-l-arginine (l-NMMA) (469), suggesting a connection between KATP channels and NO synthesis (see also Ref. 266). There is additional strong evidence that the well-known hypoxic vasodilation in the coronary system is mediated exclusively by KATP channels present in the coronary endothelium (508). Endothelial KATP channels may also contribute significantly to the pharmacological profile of KATP channel agonists. For example, in addition to their well-described effects in smooth muscle, KATP channel openers may directly act on the endothelium (528), and endothelium removal decreases the sensitivity of some arteries to KATP channel openers (395). KATP channels may also regulate the release of the vasoactive substances such as the vasoconstrictor endothelin-1 (267). Endothelial KATP channels may have a role in [Ca2+]i homeostasis, but data are conflicting. For example, some studies show that KATP channel openers cause a transient increase in intracellular Ca2+, but this increase was not temporary related to membrane hyperpolarization (478), whereas other studies with levcromakalim found no effect on intracellular Ca2+ in rabbit aortic endothelial cells (426). KATP channel block with glibenclamide was described to increase Ca2+i in pulmonary artery endothelial cells (864). The reasons for these conflicting data are unclear, and further studies are needed to clarify the mechanisms by which KATP channels affect endothelial function.
F. KATP Channels in Mitochondria
1. Biophysical and pharmacological properties of native channels
A mitochondrial KATP channel has first been described in the inner membrane of fused giant mitoplasts from rat liver mitochondria (376). The channel has a small (9pS) conductance, and is weakly blocked by ATP (IC50 of ∼800 μM) and also blocked by 4-aminopyridine and glibenclamide. The literature is not consistent, however. Other studies also reported the reconstituted channel to have a low sensitivity to MgATP block (IC50 = 335 μM), with incomplete block at 1 mM MgATP (585, 918), but with the channel having a unitary conductance of 24–105 pS (54, 585, 636, 918) (Table 4). Patch clamping of the inner mitochondrial membrane of human T-lymphocytes also demonstrated the presence of mitochondrial KATP channels (153). The reconstituted mitochondrial KATP channel is activated by superoxide, isoflurane, and diazoxide and inhibited by 5-HD and glibenclamide (54, 585, 918). Not all studies, however, are in agreement with these findings. In rat brain, for example, reconstitution of the purified inner membrane from rat brain mitochondria into a planar lipid bilayer yields mitoKATP channels with a large (∼220 pS) single-channel conductance that is not blocked by 5-hydroxydecanoic acid (113). Mitochondrial KATP channels were not observed in all electrophysiology studies (17).
Mitochondrial KATP channels are often studied nonelectrophysiologically by using surrogate assays such as K+ flux measurements, with fluorescent dyes (44, 263, 636) or with 86Rb+ fluxes (55). A recent development is the use of Tl+-sensitive fluorophores (877). In contrast to the weak effect of ATP on the channel in patch-clamp studies (see above), K+ fluxes of the reconstituted mitochondrial KATP channel in liposomes are efficiently inhibited by MgATP (IC50 of 45 μM), MgADP (IC50 of 280 μM), and glibenclamide (IC50 of 50 nM). Interestingly, Mg2+ is required for ATP inhibition, whereas glibenclamide is not effective when Mg2+ is present (636), which is a situation unlike that found for plasma/sarcolemmal KATP channels. In liposomes, the ATP-inhibited K+ fluxes can be restored by diazoxide and cromakalim. Pharmacological properties of mitochondrial KATP channels are listed in Table 3. It should be pointed out that there is not universal agreement in the literature on the presence and role of these channels; some investigators were unable to find evidence of KATP channels in mitochondria (53, 91, 159, 309). For a lively discussion on whether the existence of mitochondrial KATP channel is a fact or whether it represents fiction, please refer to a recent debate manuscript (260). For further reading on channels in the mitochondrial inner membrane, please refer to a recent review (944).
2. Molecular composition
Biochemically, 125I-glibenclamide labels a 28 kDa protein in bovine heart mitochondria (774), whereas the fluorescent probe BODIPY-glibenclamide labels a 64 kDa protein in brain mitochondria (44). The identity of these proteins remains unclear. Some reports demonstrated that conventional KATP channel subunits, Kir6.x, and SURs are present in mitochondria, but others found this not to be the case (Table 5). Likely reasons for these discrepancies include the purity of the mitochondrial fractions (with possible contamination of other intracellular membrane compartments) and the relative lack of specificity of many commercial antibodies against Kir6.x and SURx subunits. For example, immunoprecipitates from isolated heart mitochondria obtained with anti-Kir6.1 commercial antibodies yield proteins with molecular sizes of 48 and 51 kDa. MS/MS analysis of these proteins demonstrated them to be mitochondrial isocitrate dehydrogenase and NADH-dehydrogenase flavoprotein 1 (241). When expressed in HEK-293 cells, channels with a Kir6.1/SUR1 subunit combination have a similar pharmacological profile to mitochondrial KATP channels, namely, activation by diazoxide, block by 5-HD, and relative insensitivity to HMR-1098 (512). Immunofluorescence experiments, however, fail to detect Kir6.1 or Kir6.2 in mitochondria of cardiac myocytes, either after viral delivery (718) or with cardiac-specific transgenic overexpression (810). An immunocytochemistry study of isolated ventricular myocytes with a panel of antibodies against KATP channel subunits concluded that Kir6.x/SURx subunits do not localize to mitochondria (579). Moreover, mitochondrial function and mitochondrial KATP channel activity are unaffected in Kir6.1−/− or Kir6.2−/− mice (595, 876), suggesting that these subunits are not essential components of the mitochondrial channel. The current consensus appears, therefore, to be that these subunits do not comprise the mitochondrial KATP channel (718, 876, 920). Flavoprotein oxidation elicited by diazoxide (measured as an increased autofluorescence), however, is absent in ventricular myocytes from SUR2−/− mice (901), suggesting a possible role for SUR2. Indeed, low-molecular-weight SUR2 isoforms have been identified that may be components of the mitochondrial KATP channel. When heterologously expressed, a 55 kDa SUR2 splice variant colocalizes with mitoTracker dyes in COS1 cells (901). However, when coexpressed with Kir6.x, patch-clamp methods demonstrate the presence of plasmalemmal KATP channels that are blocked by millimolar ATP levels but are practically insensitive to pinacidil, diazoxide, or glibenclamide (3). Although this finding awaits confirmation, there is now a report suggesting that ROMK2, a splice variant of Kir1.1, may contribute to the molecular composition of mitochondrial KATP channels (240) (for a detailed discussion, see sect. III).
3. Physiological roles of mitochondrial KATP channels
The majority of observations concerning the function of mitochondrial KATP channel are based on modulation of mitochondrial function by K+ channel inhibitors and openers. These indexes include changes in mitochondrial matrix volume, mitochondrial potential, and oxygen consumption (170, 171, 636). The primary function of the mitochondrial KATP channel appears to be to allow K+ transport into the mitochondrial matrix which, in concert with the operation of a K+/H+ antiporter, may be involved in mitochondrial volume homeostasis (257, 258). The resulting matrix swelling may improve the rate of oxidative metabolism and the formation of reactive oxygen species. The activity of the mitochondrial KATP channel activity may also be coupled to the cellular energetic state via its inhibition by ATP and ADP (386). Nevertheless, the mitochondrial KATP channel appears to have an important function under basal conditions, which should become clearer when molecular approaches have been developed to study this channel.
G. KATP Channels in Other Cell Types That May Affect Cardiovascular Function
1. KATP channels regulate neurotransmitter release from sympathetic nerves
A novel role for KATP channels has recently been described in the sympathetic nervous system, where it was found that norepinephrine (NE) release is inhibited by active KATP channels (612). The data supporting this finding are that KATP channel openers inhibit NE release as well as the increase in atrial rate induced by electrical stimulation of the sympathetic ganglion (571, 612). In contrast, KATP channel blockers had the opposite effects. Interestingly, NE release (as well as changes in atrial rate by sympathetic nerve stimulation) were quite sensitive to diazoxide (571), which activates SUR1- or SUR2B-containing KATP channels much more effectively than cardiac SUR2A-containg KATP channels (Table 3). However, the molecular nature of these channels (as well as the signaling pathways responsible for KATP channel-regulated exocytosis) still needs to be investigated. These channels were also highly sensitive to sulfonylureas, which raises potential issues for diabetes patients treated with these compounds.
2. KATP channels regulate acetylcholine release
Acetylcholine (ACh) is synthesized and released in the central nervous system as well as from autonomic ganglia in the peripheral nervous system and in postganglionic parasympathetic neurons. In the heart, vagal nerve stimulation causes ACh release, which slows the heart rate by G protein-mediated activation of Kir3.x channels. It was found that ACh release evoked by electrical stimulation of isolated guinea pig atria was stimulated by KATP channel blockers, suggesting a role for negative feedback of KATP channels in exocytotic processes in these neurons (435). Interestingly, ACh release by neurons in the ileum is affected similarly by KATP channel activity (942), whereas mesenteric neurons are not affected by KATP channel modulation (714, 942). These findings suggest diversity of function and/or pharmacology of KATP channels in various neuronal compartments, which remains to be studied. The KATP channels in dorsal vagal neurons may be composed of Kir6.2/SUR1 subunits (425) (as in the pancreas; Table 2), which again raises issues of potential cardiovascular side effects of sulfonylurea treatment in the setting of diabetes. The study of the molecular composition of the channels in these neurons, the molecular mechanisms responsible for coupling KATP channel activity to exocytotic release, and the pathophysiological consequences of these findings await more refined tools (such as mice lacking KATP channels in specific tissue compartments).
VIII. CARDIOVASCULAR PHENOTYPES OF GENETIC MOUSE MODELS FOR STUDYING KATP CHANNELS
Gene targeting in mice has been useful to elucidate the physiological and pathophysiological functions of KATP channels. Knockout mouse models have been developed in which expression of each of the KATP channel subunits has been eliminated and some of the KATP channel subunits have been transgenically overexpressed in a tissue-specific manner.
A. Pore-Forming Subunits
The Kcnj1 gene is located on mouse chromosome 9 (32,372,418 to 32,399,192) and has three exons. Targeted disruption of exon 2 in mice leads to neonatal mortality, with ∼95% succumbing before 3 wk of age (523). The surviving mice had severe kidney defects and are dehydrated with metabolic acidosis, dysregulated blood levels of Na+ and Cl−, reduced blood pressure, polydipsia, polyuria, and poor urinary concentrating ability. With careful breeding the survival rate can be improved, but kidney defects remain (526). Heterozygous Kcnj1+/− rats have a slightly lower (2–5 mmHg) blood pressure than wild-type littermates, and a substantially lower blood pressure when on a high-salt diet (935). The latter study demonstrated an important role for the ROMK channels to maintain kidney function and revealed a critical role of ROMK channels in blood pressure regulation. No other cardiovascular phenotypes (or mitochondrial defects) have so far been reported for Kcnj1−/− animals.
In mice, the Kcnj8 gene resides on chromosome 6 at location 142,522,146 to 142,528,581 and it contains three exons. The coding region of Kir6.1 is contained in exons 2 and 3. In the Kir6.1−/− mouse, the strategy was to delete the coding region of exon 3 (565). Phenotypically, these mice develop vascular problems. They lack vascular KATP channels and develop coronary artery vasospasms that resemble Prinzmetal (or variant) angina in humans (565). Conditional knockout mice that lack Kir6.1 specifically in smooth muscle cells are hypertensive, and the vascular smooth muscle cells fail to respond to vasodilators (35), underscoring the important role of Kir6.1 in blood flow control. Hypotension was also noted in mice with transgenic expression of Kir6.1 gain-of-function subunits in vascular smooth muscle (492). Despite the significance of these findings, the Kir6.1−/− mice or vascular Kir6.1 overexpression approaches have not been employed to investigate its role of blood flow in pathophysiological conditions, such as in the context of cardiac ischemia, reperfusion, ischemic preconditioning, and the no-reflow phenomenon, where maintaining adequate blood flow is expected to beneficial. Transgenic expression of dominant-negative, pore-mutant, Kir6.1 subunits in the endothelium has revealed a role for KATP channels in the release of endothelin-1 from the endothelium (536). The release of endothelin-1 from the hearts of these mice was elevated compared with wild-type littermates, which was associated with an increased coronary resistance. Pretreatment of mice with bosentan (a blocker of ET-1 receptors) restored the elevated coronary resistance. This finding suggests that opening of endothelial KATP channels is associated with suppression of ET-1 release through cellular mechanisms that are yet to be determined. It should be noted that the latter result does not imply an exclusive role for Kir6.1 in the endothelium since the mutant subunits may heteromultimerize with endogenous Kir6.2 to eliminate KATP channel function (810, 907).
The Kcnj11 gene (encoding the Kir6.2 subunit) is located on mouse chromosome 25 (position 19,105,218 to 19,106,363). It is intronless and contains the entire coding region, which has been targeted for deletion (564). Since Kir6.2 forms a component of the pancreatic β-cell KATP channel, which triggers insulin release, one would expect these mice to be diabetic. However, the Kir6.2−/− mice largely lack the expected hyperglycemic phenotype, but have been used to demonstrate that Kir6.2 is a key regulator of glucose- and sulfonylurea-induced insulin secretion (720).
Mice with cardiac-specific transgenic overexpression of Kir6.2 gain-of-function mutant subunits (Kir6.2-ΔN30,K185Q) exhibited no gross physiological or morphological phenotypes (235, 458). Interestingly, however, although the ATP sensitivity of the KATP channels was decreased in excised patches, channels remained closed in the intact cell. A similar observation was made with cardiac-specific expression of a Kir6.2 subunit carrying a mutation associated with neonatal diabetes (Kir6.2-V59M mutant), which led to channels with altered ATP sensitivity, but no cardiac abnormalities (120). Transgenic overexpression of SUR1 and gain-of-function Kir6.2 (ΔN30, K185Q) subunits in the ventricle results in embryonic lethality (807), but the physiological relevance of this overexpression study is not clear. Knockout approaches have also been used. The Kir6.2−/− mice, which lack ventricular KATP channels, have no overt obvious cardiac defects (720). Closer inspection has revealed important physiological roles for KATP channels in the cardiovascular system. The first demonstration was that the Kir6.2−/− mice had an impaired tolerance to exercise and stress (940). The fact that it was the ventricular KATP channel (and not the loss of KATP channels in another somatic compartment) that was responsible for this deficit was underscored by the use of transgenic mice with cardiac-specific overexpression of dominant-negative Kir6 subunits, which likewise had an impaired exercise tolerance (536). Kir6.2−/− mice also develop arrhythmias and sudden death with sympathetic challenge (likely due to cellular Ca2+ overload since it is preventable by Ca2+ channel blockers), which points to a vital protective function of KATP channels under physiological conditions (940). A recent finding points to a more subtle and important physiological function: when the heart rate is elevated, the action potential duration adapts (decreases) over the time course of a few minutes (202), which is responsible for a decreased refractoriness and prevention of arrhythmias. Both glibenclamide and loss of KATP channel function (achieved by cardiac-specific overexpression of dominant-negative Kir6 subunits) largely eliminates frequency adaptation, which points to a key physiological action of KATP channels in action potential adaptation at elevated heart rates (941). Moreover, exercise led to upregulation of ventricular KATP channel subunit expression as well as a more robust KATP channel-dependent heart rate adaptation in mice (941). The Kir6.2−/− mouse model is useful to study sarcolemmal KATP channels, since ventricular KATP channels are absent in the mice, whereas mitochondria KATP channel function is preserved (771). Indeed, they have helped to elucidate important roles for KATP channels during pathological processes such as ischemia and ischemic preconditioning (see sect. IX).
B. Accessory Subunits
The SUR1 subunit contains 17 transmembrane segments and is encoded by the Abcc8 gene on the mouse chromosome 25 (at location 19,072,876 to 19,091,318). Abcc8 is comprised of multiple exons, and transcriptional activity can give rise to several SUR1 splice variants (289, 323, 330, 685, 704). Two groups have produced SUR1−/− mice by either deleting exon 1, containing promoter regions and the starting methionine (732), or exon 2 (717). Although neither mouse model displayed the expected severe dysregulation of insulin release, they have been useful in helping us to understand how insulin release is fine-tuned (5). SUR1 also has other functions, but reports demonstrating a role in the cardiovascular system are only now emerging. For example, the SUR1−/− mice were instrumental in the identification of a novel subtype of SUR1-based, diazoxide-sensitive, KATP channels in mouse atria (237), which may also be present in the human atria and ventricle (222). The SUR1-null mice are unexpectedly protected from cardiac ischemia; they exhibit a reduced postischemic infarct size and preservation of left ventricular function upon reperfusion after ischemia (207). Interestingly, the spread of the hemorrhagic contusion and capillary fragmentation in spinal cord injury are also prevented in SUR−/− mice (745). These results therefore suggest a detrimental role for SUR1 during stress conditions. It should be noted that SUR1 effects may not originate in the cardiovascular system itself, but it may affect the sequelae of ischemia/reperfusion through potential actions on other cell and tissue types (487).
The SUR2 subunit is coded by the Abcc9 gene on mouse chromosome 6 (position 142,546,356 to 142,659,537) and has 40 exons (2 of which are untranslated). Several splice variants have been identified (114, 164, 727, 901). The two “full-length” splice variants that are most commonly studied, SUR2A and SUR2B, differ from each other by alternative use of exons 39 and 40. SUR2−/− mice have been generated in which two exons were targeted to remove regions in the first of the two intracellular nucleotide binding folds that are involved in ATP/ADP binding (116). Similar to the Kir6.1−/− mice, these mice are hypertensive and exhibit spontaneous coronary vasospasms (115). The SUR2−/− mice are paradoxically protected against ischemia (762). The protection may be due to spontaneous coronary vasospasms that occur in the SUR2−/− mouse hearts that may cause short bursts of ischemia (similar to ischemic preconditioning). Restoring SUR2 specifically in the heart by transgenic expression within the SUR2-null background led to further protection against ischemia-induced infarct development and improved postischemic contractile recovery (763). Another study also demonstrated that nonspecific transgenic overexpression of SUR2A in mice (driven by the CMV promoter) is protective against ischemia and that the hearts from these mice had reduced infarct sizes in response to ischemia-reperfusion compared with wild-type (189). The interpretation of the data with the SUR2−/− mice is complicated due to the presence of remaining short-form SUR2 isoforms (727). A different knockout design (deleting exon 5) that eliminates all SUR2 isoforms has recently been described. Homozygous exon 5 deletion (SUR2-Ex5) results in neonatal mortality with progressive cardiac dysfunction in the first weeks of life (212). At present it is unclear why SUR2 knockout results in such a severe phenotype, whereas Kir6.2 knockout mice have milder cardiovascular defects.
IX. PATHOPHYSIOLOGICAL ROLES OF KATP CHANNELS
Some of the physiological roles of cardiovascular KATP channels have been highlighted in the section VII and include functions such as the frequency adaptation of the ventricular action potential duration, the release of atrial natriuretic peptide from the atria, vasodilation and increased blood flow in smooth muscle, release of vasoactive substances from the endothelium, and mitochondrial matrix volume regulation. It is conceivable that any of these functions may impact cardiovascular pathophysiology, but few of these have been studied in detail. In this section we deal mainly with stress responses and the consequences of coronary artery disease.
A. Cardiac KATP Channels Protect Against Stress
Mice with somatic deficiency of Kir6.2 subunits or with transgenic, cardiac-specific overexpression of dominant negative Kir6 subunits have a phenotype of impaired stress responses and intolerance to an exercise load (described in sect. VIII). Cellular mechanisms are likely to relate (at least in part) to the prevention of intracellular Ca2+ overload (Figure 12). Indeed, upon catecholamine challenge, ventricular myocytes isolated from the hearts of Kir6.2−/− mice exhibit defective action potential shortening, which predisposes the myocardium to early afterdepolarizations and arrhythmias (510). In further support, metabolic inhibition with the mitochondrial uncoupler DNP leads to Ca2+ overload in COS-7 cells, which is mitigated when cells are transfected with Kir6.2/SUR2A cDNAs (408). There is in vivo evidence for a central role of KATP channels in protecting the myocardium against Ca2+ overload: using manganese-enhanced cardiac magnetic resonance imaging, myocardial Ca2+ accumulation was exacerbated and myocardial function was impaired in hearts of Kir6.2−/− mice (305). Cardiac overexpression of ATP-insensitive Kir6.2 mutant subunits also results in defects in intracellular Ca2+ handling (655). In addition to the effects on intracellular Ca2+, KATP channels may additionally protect against stress by preserving mitochondrial function. In rat ventricular myocytes, for example, shRNA knockdown of Kir6.2 (by 50%) not only leads to disrupted intracellular Ca2+ homeostasis, but also to oscillations of mitochondrial membrane potential (764). Mitochondrial KATP channels may of course also directly participate in the protection against stress (259). Targeted expression of Kir6.2 in the mitochondria of HEK293 and HL-1 cells confers protection against hypoxic stress (516). Although the role of Kir6.2 in mitochondria is questionable (see Table 5), this study provides good nonpharmacological evidence that increased K+ fluxes in mitochondria may participate in protection against stress. Finally, KATP channels may facilitate the cardiac response to stress by regulating PGC-1α and its target genes, in part through the FOXO1 pathway (366).
B. Myocardial Ischemia and Reperfusion
The leading cause of heart disease and heart failure is a deficit of coronary blood flow due to atherosclerosis and thrombosis. The reader is referred to reviews and books for full details on this subject (94, 204, 243, 339, 396, 397, 623, 630). In short, myocardial ischemia occurs when blood and oxygen supply do not meet myocardial demand. The lack of oxygen and nutrients supply and the accumulation of metabolic waste products result in a progressive succession of events that starts with immediate biochemical and functional abnormalities, including metabolic impairment, rapid K+ loss from myocytes within the ischemic zone, and extracellular K+ accumulation. Loss of myocardial contractility occurs within 60 s and other changes that take a more protracted time course. Reperfusion (restoration of blood flow) can occur, but loss of viability (irreversible injury) occurs within 20–40 min following total occlusion of blood flow. The loss of contractility and the development of lethal arrhythmias are responsible for the majority of sudden death events. As summarized below, KATP channels appear to have a role in many of these processes [also refer to other reviews (69, 236, 293, 295, 338, 611, 622)]. Cardiac ischemia is a complex event, and the effects of KATP channel opening are more complex. Overall, KATP channel opening can be both detrimental (e.g., by precipitating arrhythmias and maintaining ventricular fibrillation) and protective (e.g., by limiting he development of a post-ischemic infarct).
1. K+ efflux from ischemic tissue: do KATP channels contribute?
The loss of K+ ions from the ischemic myocardium is an early response and starts to occur within seconds after the interruption of oxygenated blood flow (328, 351, 445). Due to flow restriction, K+ accumulates in the extracellular space, which in turn causes membrane depolarization, action potential shortening, and a decreased rate of rise of the action potential (which is an important determinant of rapid electrical conduction). Initially, the hypothesis was raised that K+ efflux occurs as a result of an ionic shift that occurs in response to electrogenic anion loss from ischemic cells in the form of lactate (137, 443). The finding that hypoxia and metabolic inhibition result in an increase of a specific K+ conductance associated with 42K+ efflux (379, 847, 850), coupled with the description of the KATP channel (607, 815), led to studies to investigate the role of this channel during the initial phase of K+ efflux. In the earliest studies performed with isolated rat, guinea pig, and rabbit hearts rendered globally ischemic by stop-flow or low flow, glibenclamide (1–10 μM) was found to strongly attenuate extracellular K+ accumulation during ischemia (424, 871). Moreover, there was no correlation between early lactate and K+ efflux rates (424), and glibenclamide had no effect on the extracellular acidification (871). Using a variety of experimental preparations, others have also reported glibenclamide, more so than tolbutamide, to inhibit K+ or Rb+ loss during simulated myocardial ischemia (399, 835, 866, 870). The KATP channel blocker 5-hydroxydecanoate also inhibits K+ release from the ischemic rat (600) or guinea pig (681) myocardium. Thus KATP channels appear to have a role in K+ loss early during cardiac ischemia. It should be noted, however, that glibenclamide only partially prevents K+ accumulation and is only effective during the early (∼10 min) phase of ischemia (424, 871). Moreover, genetic deficiency of KATP channels (experiments with Kir6.2−/− mice) does not prevent K+ accumulation during ischemia (680). Therefore, in addition to KATP channels, other mechanisms contribute to K+ accumulation in the ischemic myocardium, including shrinkage of the extracellular space, diminished K+ reuptake due to Na+/K+ pump inhibition, and K+ loss through other pathways, such as inward rectifier K+ channels, Na+-activated K+ channels, arachidonic acid-activated K+ channels, and other mechanisms (273, 670, 800, 870). It is also possible that K+ loss occurs as a charge balance mechanism to compensate for Na+ influx that occurs during ischemia. Indeed, in the presence of blockers of Na+ influx pathways (Na+ channels, L-type Ca2+ channels, Na+-K+-2Cl− cotransport, and the Na+-H+ exchange), tissue K+ accumulation in response to ischemia is completely prevented in an arterially perfused rabbit interventricular septum preparation (733). In this paradigm, Na+ influx into ischemic cells is the major driver, and the KATP channel (along with other K+ permeation pathways) acts merely as a flux coupling pathway to maintain electroneutrality in the ischemic tissue. This mechanism may also explain why KATP channel openers (such as cromakalim) do not lead to K+ accumulation in normoxic tissue, why KATP channel blockers only partially prevent K+ accumulation, and why in some studies (but see Ref. 569) KATP channel openers (such as cromakalim) fail to enhance K+ loss during ischemia (424, 733, 843).
2. Role of KATP channels during action potential shortening that occurs during acute ischemia
Action potential shortening occurs during hypoxia and metabolic inhibition. KATP channel opening is involved since it can be prevented by glibenclamide (177) or genetic knockout of Kir6.2 subunits (270). Following coronary artery occlusion, action potential duration shortening also occurs (associated with diastolic membrane potential depolarization) (392, 445, 483), but the role of the KATP channel may be quite different since the APD shortening during ischemia happens for different reasons (due to K+ accumulation in the restricted extracellular spaces caused by inadequate blood flow and washout; see above). Indeed, action potential duration shortening occurs with the same time course as extracellular K+ accumulation, starting within 10–30 s of coronary occlusion and progressing over the next 10–15 min (445) and then stabilizes to some extent (615). Moreover, in dog heart, infusing K+ to raise the extracellular K+ from 3.4 to 5.9 mM causes action potential shortening similar to that observed with ischemia, and superimposed hypoxia does not lead to significant additional shortening (183). Conduction slowing, in contrast, cannot solely be explained by raising the extracellular K+ (183, 446), and conduction paradoxically even accelerates transiently during the first minute of ischemia due to the elevated extracellular K+ levels (205, 243). Intrinsic changes in ionic currents also contribute to the action potential changes. Experimental data and modeling studies support a role for a decreased Na+ current as being responsible for conduction slowing (99, 726) with an additional role for the L-type Ca2+ current in action potential shortening (438). KATP channel activation may also contribute to APD shortening. However, since KATP channels contribute only partially to extracellular K+ accumulation during ischemia, which is the major determinant of action potential shortening and membrane depolarization, modulators of KATP channels are expected to only have partial effects. Indeed, blocking KATP channels with glibenclamide or 5-hydroxydecanoate slows (but does not prevent) action potential shortening during ischemia in a variety of experimental models and in different species (Table 6). Moreover, most studies found that KATP channel openers, including levcromakalim, pinacidil, HMR-1098, and diazoxide, accelerate the ischemia-induced action potential shortening (Table 6), at least within the first 10–15 min before a plateau is reached (615).
3. Reasons for KATP channel open during ischemia
The question appears therefore not to be whether KATP channels open during ischemia, but what the reasons are for their opening, considering that ATP needs to decrease to submillimolar levels for them to open (at least in patch-clamp experiments) (124). This finding appears inconsistent with the rapid CrP depletion, but slower decline in tissue ATP levels that occurs during myocardial ischemia (211, 465, 574). Experimental data demonstrate that a poor relationship exists between cellular ATP levels and KATP channel opening. For example, during metabolic blockade of ferret hearts, severe action potential shortening occurs (44% after 5 min) at a time when there were no major changes in the tissue ATP concentration (it decreased from 5.4 to 4.3 mM) (206). Subsequent addition of glucose led to a 70% recovery of APD with only a 10% increase in ATP concentration. Also, in voltage-clamped rabbit papillary muscles, KATP channel-dependent APD shortening occurs during hypoxia, despite only a modest (∼25%) decline in tissue ATP content (177). A contributing factor might be the sharp and rapid decline in the calculated free energy of ATP hydrolysis during the first few minutes of ischemia (233). Other factors occurring during ischemia, such as cellular acidosis, changes in inorganic phosphate, adenosine production, and changes in phospholipid content may act to lower the channel′s ATP sensitivity and/or to promote its opening (124). For example, the KATP channel is activated by AMPK (767, 905), the guardian of cardiac energy status (325), and by adenylate kinase phosphotransfer reactions (100), as a result of small changes in intracellular AMP, which occurs rapidly. In dog heart, for example, the AMP/ATP ratio doubles within just 1 min of regional ischemia, mainly due to an increase in the AMP levels (621). Other factors may further contribute to KATP channel opening. For example, due to their high surface density and large unitary conductance, significant electrophysiological alterations are possible when only a small percentage of the KATP channels open (232, 726). There may also be intracellular factors that remain to be identified that may modulate KATP channel opening (786). Finally, it may not be the total cellular ATP/ADP levels that regulate KATP channels, but rather the levels in the immediate microenvironment of the channel. Indeed, O2 and ATP gradients are known to occur inside cells (407). The local ATP level at the channel will be determined by the site and rate of ATP production, the rate of ATP diffusion, and the rate of ATP consumption. The source of ATP may be more important than cellular content as a cause of intracellular ATP compartmentalization (624). In this regard, KATP channels are preferentially regulated by glycolysis (178, 179, 866) and directly associate with glycolytic enzymes (141, 179, 358, 409, 433), which lends credence to the idea that KATP channel activity is regulated locally by compartmentalized changes in nucleotide concentrations.
4. Role of KATP channels in arrhythmias that occur during ischemia and reperfusion
Ventricular arrhythmias can occur early (first 10 min) during ischemia (phase 1a arrhythmias), or the occurrence can be delayed. The phase 1b arrhythmias occur between 15 and 60 min after the onset of ischemia, and phase 2 arrhythmias occur even later (after ∼90 min) (394). Serious arrhythmias can also ensue upon reperfusion of the ischemic tissue. Distinct mechanisms underlie these types of arrhythmias (123, 444, 484, 676), and KATP channel opening may not necessarily affect these types of arrhythmias in the same way (if at all).
a) initiation of arrhythmias early during cardiac ischemia. Early during ischemia, the membrane potential depolarizes and action potential shortening occurs, but only within the ischemic zone (due to extracellular K+ accumulation; see above). A key role for K+ is illustrated by the finding that an increased K+ concentration in the perfusate (to balance the K+ gradients) prevents arrhythmias during early ischemia (but not during reperfusion) in isolated rat hearts (527, 679). Since the electrophysiological changes are regional, gradients are established between the ischemic and the nonischemic zones, thus creating an electrical border zone in which arrhythmias can ensue (95, 205, 391). During this early stage, electrical gradients and dispersed refractoriness, together with slowing of conduction, contribute to the formation and maintenance of the ischemic arrhythmias (329, 446, 865). Arrhythmias can be triggered by extrasystoles (“triggered activity”) or by abnormal automaticity. The “trigger” may take the form of an afterdepolarization occurring during the action potential (“early afterpotentials”) or during diastole between action potentials (“late afterpotentials”). Given this complex set of events, it is difficult to predict how KATP channel opening may affect the outcome. In general, triggered activity and automaticity are promoted by interventions that oppose repolarization or hyperpolarization, such as an increased inward current or a decreased outward current. Conversely, larger outward currents (such as when KATP channels open) should mitigate these arrhythmogenic triggers. An argument could therefore be made that KATP channel opening may be anti-arrhythmic since the outward current generated will oppose these arrhythmogenic triggers. Indeed, in cellular assays, KATP channel openers inhibit experimentally induced abnormal automaticity (481, 505, 753), or triggered arrhythmias elicited by early afterpotentials (234, 753) or late afterpotentials (481, 505, 753). A counter argument, however, is that KATP channel opening may be pro-arrhythmic (at least to some extent) since they contribute partially to the primary defect, which is the loss of K+ from ischemic cells and K+ accumulation in the ischemic zone. Both of these arguments probably hold elements of the truth, given the spectrum of experimental outcomes when studying the effects of KATP channel openers and blockers on arrhythmias during ischemia (Table 7). On balance, it appears that KATP channel opening is detrimental during ischemia by causing arrhythmias. KATP channel opening (for example in the peri-infarct area) is expected to reduce the APD to cause heterogeneity in repolarization, which is a substrate for reentrant arrhythmias (68, 869). This may be an explanation for the antiarrhythmic effects of KATP channel blockers ischemia (869), both in experimental models (424, 872) and in patients with transient myocardial ischemia or acute myocardial infarction (97, 519).
b) electrical conduction changes during ischemia and the specialized cardiac conduction system. Electrical conduction is transiently increased within the first few minutes after ischemia (354) (due to the increases in extracellular K+) before it rapidly and progressively decreases (40, 73, 205, 394). A major cause for conduction slowing during early ischemia is depolarization-induced reduction in the action potential (AP) amplitude and upstroke velocity. Conduction slowing occurs both in the working myocardium and in the subendocardial layers that accommodate the specialized CCS (394), which consists of specialized myocytes of the AV node, bundle of His, free-running Purkinje strands, and a vast network of subendocardial Purkinje fibers. This cellular network provides rapid electrical propagation from the AV node to the ventricular myocytes and ensures synchronous excitation and contraction of the entire heart. Many ventricular arrhythmias initiate in the Purkinje fiber network, mostly through local reentrant circuits (79). During ischemia, small changes in conduction velocity, refractory period, or conduction circuit path length contribute to the initiation of reentrant circuits and life-threatening arrhythmias. The AV node and distal Purkinje network are particularly sensitive to even short periods of ischemia (41). Consistently, the presence of an intact Purkinje network is a necessary component for the initialization of VF during acute regional ischemia (393). Understanding the properties of the CCS during ischemia potentially offers therapeutic opportunities for the treatment of this type of arrhythmia. KATP channels appear to have some role. KATP channels are expressed in the SA node and CCS and affect the functional properties of these cells during metabolic stress (415, 502, 696). For example, hypoxia-induced bradycardia is mediated by KATP channel opening (248, 498). Moreover, hypoxia-induced conduction slowing in Langendorff-perfused rabbit hearts is prevented by the KATP channel blocker glibenclamide (698). In dog heart, intramyocardial conduction delay induced by ischemia is reduced by glibenclamide (61). Glibenclamide also decreases the incidence of ischemia-induced transmural conduction block in a canine wedge preparation (633), prevents the increased longitudinal resistance (electrical uncoupling) in an isolated rabbit septal preparation during ischemia (787), and prevents the subendocardial conduction delay produced by ischemia in an isolated, arterially perfused canine interventricular septal preparation (576). CCS KATP channels are therefore likely to participate in ischemia-induced conduction disturbances, and their opening may precipitate life-threatening arrhythmias. The situation may be aggravated by the regional distribution of KATP channels. For example, KATP channels may contribute to the left-right ventricular electrical heterogeneity in guinea pig hearts during global ischemia [but not in the dog heart (797)], mainly due to the higher KATP channel density in the left ventricle (629). Simulation studies further predict that transmural gradients of KATP channel density may contribute to activation rate gradients and arrhythmogenicity, with a potential role for the Purkinje system in causing local endocardial spatial heterogeneity, which may serve as an arrhythmogenic trigger (80).
c) termination of arrhythmias by blocking KATP channel activity. From the preceding paragraph, it is clear that KATP channel opening is a potential arrhythmogenic substrate. Experimental studies are underway to explore the underlying arrhythmogenic mechanisms of KATP channel opening (68). There are several studies to show that blocking KATP channels with glibenclamide or 5-HD decreases the incidence of ischemia-induced arrhythmias or to have an antifibrillatory effect by decreasing the duration of ventricular fibrillation (Table 7). This antiarrhythmic effect is exacerbated by the KATP channel-dependent action potential duration dispersion and refractoriness between the normal and ischemic tissues. Regional dispersion has also been noted to occur between the left and right ventricles in guinea pig heart (629), but not in the Langendorff-perfused ischemic dog heart (797). In nonischemic, cardiomyopathic Langendorff-perfused human hearts, glibenclamide was found to terminate ventricular fibrillation (VF) (218), leading to the concept that KATP channel opening participates in the maintenance of VF. They may do so by stabilizing rotor dynamics during VF, as suggested from a study performed in pigs (650). Thus the stage is set for future studies to determine whether KATP channel block may represent a viable therapeutic antiarrhythmic target.
5. KATP channel also protect against injury during ischemia and reperfusion
Upon coronary occlusion, cardiac ischemia not only results in K+ efflux and the electrophysiological changes discussed above. Additionally, among others, cardiac function rapidly deteriorates, intracellular pH falls, biochemical changes take place (e.g., inorganic phosphate levels increase, CrP levels decrease, and lactate is formed), changes occur in lipid metabolism, and oxygen free radical formation occurs. During reperfusion after short periods of acute ischemia, contractile function recovers (at least partially). Ischemia ultimately results in cellular necrosis and the formation of an infarct. The severity of ischemia is often measured by the rate and magnitude of postischemic contractile recovery (i.e., during reperfusion) and by the size of the infarct that forms. KATP channels affect both. This is a potentially important clinical finding, given that the major determinant of the long-term prognosis in patients that survive an ischemic event is the size of the infarct. Understanding the mechanisms involved in cardioprotection and infarct size development may therefore have important therapeutic benefits.
a) protection against contractile dysfunction during reperfusion. A large number of studies have been performed to examine the anti-ischemic effects of KATP channel openers (338, 622). The overwhelming majority of the studies demonstrate that KCOs delay the onset of rigor contracture during ischemia and improve contractile recovery of the heart during postischemic reperfusion (Table 8). The protective effect ranges over a variety of species, from rodents to dogs, and appears to be independent of the experimental preparations used (in vivo, isolated heart preparations and in vitro preparations). The anti-ischemic effect is evident with structurally diverse compounds, including nicorandil, aprikalim, cromakalim, levcromakalim, diazoxide, HMR 1883, P-1075, Y-26763, and pinacidil. In the vast majority of the studies, the protective effects of KCOs are absent (or mitigated) in the presence of KATP channel blockers (glibenclamide or 5-hydroxydecanoate), suggesting that the protective effects are mediated by KATP channels. In most of the studies, the KCOs were applied before the onset of ischemia. When applied after the ischemic period (i.e., when applied only during the reperfusion period), the KCOs had no beneficial effect to improve postischemic contractile recovery (32, 690). Other indexes of ischemic damage were also investigated in some of the studies. For example, cromakalim and P-1075 reduced lactate dehydrogenase (LDH) and creatine kinase (CK) release during postischemic reperfusion (300, 301, 690, 782) or decreased lactate excretion in the coronary effluent (730). When applied during the ischemic period, the KCOs also have an ATP-sparing effect (see sect. IXB8). Compounds with KATP channel blocking activity, such as glibenclamide, 5-hydroxydecanoate, HMR 1098, or tolbutamide, when applied alone (in the absence of KCOs) exacerbate ischemic damage, evident as a more severe ischemic contracture and worsening of postischemic contractile recovery in most, but not all, studies (32, 133, 251, 272, 300, 336, 566, 897). Thus a large body of literature suggests that opening KATP channels during ischemia is cardioprotective.
b) protection against the development of an infarct. The extent of an infarct after acute coronary artery occlusion is dependent on the degree of ischemic damage. The anti-ischemic effects of KATP channel opening would therefore predict a reduction in tissue damage and decreased infarct development after a cardiac ischemic episode. Indeed, KATP channel openers have been demonstrated to reduce postischemic infarct development (e.g., in dogs and rabbits; Table 9). Infarct size reduction was observed with diverse agents, including cromakalim, aprikalim, bimakalim, pinacidil, and nicorandil. The cardioprotective effects of KCOs can be blocked by the KATP channel blockers glibenclamide and 5-hydroxydecanoate. By themselves (in the absence of KCOs), these KATP channel blockers mostly have little effect on infarct size development (Table 9). Not all studies are positive, however, and it appears that the choice of anesthetic used for the in vivo studies may affect the cardioprotective properties of KATP channel opening (567, 584, 854). In the majority of the studies, KATP channel openers were applied before the onset of ischemia. A few studies examined the timing of drug application and found that when KCOs were applied after ischemia (i.e., only during the reperfusion period), no benefit was observed (28, 29, 297). Very few studies examined the application of KCOs after the onset of ischemia, but before reperfusion. In such studies, when nicorandil or bimakalim was applied 10–15 min after induction of ischemia, they were able to reduce infarct size (209, 294, 475, 900). Overall, the majority of studies demonstrate that KATP channels have an important role in the protection of the myocardium from ischemia-reperfusion injury.
c) protection against the decline of cellular high-energy phosphates during ischemia. During the first few minutes of acute myocardial ischemia, a rapid loss of intracellular creatine phosphate levels occurs, which is followed by a more gradual (over the next 5–10 min) decline of ATP levels and an increase of ADP and AMP levels (233, 375). The anti-ischemic effects of KATP channel opening may translate to preservation of high-energy phosphate compounds during ischemia. Indeed, during ischemia in a guinea pig right ventricle preparation, pinacidil was found to preserve both ATP and PCr levels in a glibenclamide-dependent manner (556). The cardioprotective effects of cromakalim are also accompanied by an attenuation of intracellular ATP loss during ischemia (301). Moreover, in 31P-NMR studies, nicorandil and diazoxide were shown to mitigate ischemia-induced ATP depletion and Pi accumulation and to reduce intracellular Na+ accumulation during ischemia (182, 247, 852). Total myocardial metabolite levels during ischemia are also preserved by nicorandil (640; but see Ref. 566). Not all studies are in agreement. Other studies found that, although pinacidil and diazoxide protected against postischemic functional decline and improved ATP and phosphocreatine levels during reperfusion, no appreciable effects were observed on ischemic ATP levels (11, 239, 247, 605). Even if ATP levels are not substantially changed, KATP channel openers can improve the free energy of ATP hydrolysis, as demonstrated with 31P-NMR measurements in hypoxic guinea pig hearts (173). It is possible that KATP channel openers may preserve mitochondrial function during ischemia, similar to the preservation of mitochondrial integrity that is observed with L-type Ca2+ channel blockers (464, 591) and with blockers of the mitochondrial permeability transition pore (180), or by inhibiting Ca2+ uptake into mitochondria (224, 254). This putative protective effect may be direct, due to effects of the KCOs on mitochondrial KATP channels (263) or on the mitochondrial F1F0-ATPase (11). Protective effects of KCOs on mitochondria may also be indirect, for example, by reducing Ca2+ uptake into cardiomyocytes (and presumably Ca2+-induced mitochondrial damage) (142, 480, 764).
d) do species differences exist? An argument could be raised that there are major differences in mice and rats compared with other, larger animals. This may be so because rodents have a short cardiac cycle length and they need large outward currents to repolarize the action potential and to avoid fusion of the action potentials. Intuitively, one may expect rodents to be extremely dependent sarcolemmal KATP channels to survive metabolic stress and for glibenclamide to be devastating to ischemic mouse hearts (with much less effect of the drug in species with long action potentials). It is therefore instructive to compare the effects of glibenclamide on ischemic contracture and cardiac functional recovery and infarct size after myocardial ischemia in species with short (mouse and rat) and long (rabbit and dog) action potentials. The published data are mixed (Table 9) and do not appear to support the notion that the intrinsic action potential duration fundamentally changes in the outcome. In dog, for example, infarct size is either increased or unaffected by glibenclamide (31, 292). In some rabbit studies, glibenclamide increased postischemic infarct size (567, 805), whereas other studies reported no effect (812, 813, 854). In rats, glibenclamide has no effect on infarct size (152, 647), increases infarct size at higher concentrations that are associated with changes in blood glucose levels (511), or may even decrease infarct size in some studies (203). Postischemic functional recovery is largely unaffected by glibenclamide in mice and rats (653, 722, 723). Contrary to the notion that KATP channel block is devastating to ischemic mouse hearts, infarct size and/or functional recovery after ischemia are not markedly increased in Kir6.2 knockout mice (which lack cardiac sarcolemmal KATP channels) compared with wild-type mice (771, 856). Collectively, these data indicate that species with an inherently short cardiac action potential (such as mice and rats) are not more dependent on KATP channels to protect against stress than other species with inherently longer action potentials (such as rabbits, dogs, and perhaps humans).
6. Preconditioning of the myocardium as a protective mechanism
One or more brief ischemic episodes protect the heart from a subsequent sustained ischemic insult (582). This endogenous form of cardioprotection, named ischemic preconditioning, has been the subject of many investigations in an attempt to elucidate the underlying mechanisms, which (if understood) have potential therapeutic benefits for the treatment or prevention of myocardial ischemia and infarction. The reader is referred to reviews on the subject for a full description of preconditioning, the underlying mechanisms, and cellular pathways involved (130, 581, 756, 895, 903). Preconditioning can also be elicited by pharmacological compounds or receptor agonists (such as adenosine) (507), and this type of protection is often referred to as “pharmacological preconditioning.” Volatile anesthetics also have a preconditioning-like effect on the ischemic heart (772). The early phase of preconditioning occurs 1–3 h after the conditioning stimulus, but there is also a delayed phase, which occurs after 18–24 h and can continue for as long as 24–72 h (757). KATP channels appear to have an important role in most, if not all, in the cardioprotection resulting from these forms of preconditioning. Table 10 lists a small selection of a very large body of literature to study the effects of KATP channel modulating agents on preconditioning. Some of the earliest mechanistic studies found that ischemic preconditioning slows energy metabolism during the subsequent sustained (or index) ischemic period (583). Soon afterwards, signaling through adenosine receptors was found to mediate the protective effect of ischemic preconditioning (507). The description that blockade of KATP channels with glibenclamide prevented myocardial preconditioning in dogs (292) set in motion an avalanche of studies to examine the role(s) of KATP channel modulators in various forms of preconditioning. Initially, conflicting data appeared from different laboratories (133, 300, 442, 805). The negative reports in some studies might have been due partly to off-target effects of the KCOs (e.g., coronary steal), but it was soon appreciated that the choice of anesthetics used also play a major role (854). Today, with hundreds of citations in Pubmed on this subject, the consensus is that certain compounds with KATP channel blocking activity such as glibenclamide and 5-hydroxydecanoate mitigate the protective effects of ischemic preconditioning on infarct size, whereas others such as HMR-1098 (at least in some studies) are less effective against early ischemic preconditioning (Table 10). Glibenclamide, 5-hydroxydecanoate, and HMR-1098 also block the protective effects of late preconditioning (65, 559, 635, 811). The converse is also true: structurally diverse compounds with KATP channel opening activity, including aprikalim, bimakalim, pinacidil, nicorandil, or diazoxide, are cardioprotective by reducing the size of the developing infarct when they are applied before index ischemia (i.e., given instead of the preconditioning ischemia). The protective effects of these compounds are only evident when applied before (or early during) the index ischemia, and these protective effects can be reversed by coadministration of compounds with KATP channel blocking activity (Table 10).
7. Interpretation of the data: which KATP channels are involved in cardioprotection?
In the majority of studies, KATP channel openers are cardioprotective during cardiac ischemia (Tables 8–10). While KATP channel blockers are not necessarily detrimental, they generally abolish the protective effects of the KATP channel openers. Although the data are indisputable, the interpretation of these data has given rise to conflicting views, mostly related to the nature of the KATP channel subtype that is responsible for ischemic cardioprotection. A variety of KATP channel subtypes exist in the cardiovascular system (see sect. VII), each with unique functions in various regions of the heart and specialized conduction system, smooth muscle, and endothelium. Moreover, these channels have overlapping pharmacological profiles (Table 3). Despite this diversity, most studies attempt to discriminate between the protective roles of sarcolemmal and mitochondrial KATP channels in the ventricular myocyte. Although the arguments can be complex, the key arguments in the debate of roles for/against these channels can be summarized as follows.
a) diazoxide is a specific mitochondrial KATP channel opener. There are >100 studies in the cardiac preconditioning literature that employed diazoxide as a specific opener of mitochondrial KATP channels and/or used 5-HD as a specific mitochondrial KATP channel blocker. Indeed, in isolated mitochondria, diazoxide opens the mitochondrial KATP channel with an EC50 of 2–27 μM (263, 512, 876) and has little effect on ventricular sarcolemmal KATP channels under basal conditions. The potent cardioprotective properties of diazoxide are therefore often taken as evidence for a contribution of mitochondrial KATP channels in cardioprotection. Counterarguments can be leveled, including the fact that ventricular sarcolemmal KATP channels become sensitive to diazoxide under metabolically impaired conditions, as occurs during myocardial ischemia (151, 546), as evidenced by the fact that the rate of action potential duration shortening during the first 10 min of ischemia is accelerated by the compound (at least in some studies; Table 6). The interpretation of data obtained with diazoxide is therefore not straightforward. The fact is that diazoxide neither exhibits specificity (the capacity of a drug to have a unique action) since it affects blood pressure and blood glucose levels as well as contributes to cardioprotection, nor selectivity (the ability of a drug to affect a particular molecular target in preference to others) since it affects several types of KATP channels and also have other off-target effects (125). Moreover, diazoxide affects these various effectors within the range of concentrations that are commonly used in cardioprotection studies (125). A strong argument in favor of the sarcolemmal KATP channel as a diazoxide target comes from the observation that the cardioprotective effect of diazoxide is blunted or lost in mice lacking Kir6.2, a known subunit of the sarcolemmal KATP channel (770, 876). Assigning a single molecular target to diazoxide′s protective effect is therefore not a simple matter, and the possibility must be considered that diazoxide has multiple effectors that may synergistically contribute to this compound′s powerful cardioprotective properties (125).
b) 5-hd selectively blocks specific types of KATP channels. 5-HD is often employed in cardioprotection studies as a specific mitochondrial KATP channel blocker (43, 780). This contention appears to rely mainly of the ability of 5-HD to reverse diazoxide-induced K+ fluxes in reconstituted heart mitochondrial preparations (262) and to attenuate diazoxide-induced changes in autofluorescence of isolated cardiomyocytes (513, 693). This argument ignores the fact that 5-HD blocks (with an IC50 of 0.16–28 μM) KATP channels recorded in the inside-out configuration in isolated patches from cardiac ventricular myocytes (497, 609). 5-HD also effectively blocks ventricular KATP channel currents in the whole cell configuration (608). Moreover, as one would expect from its ability to block KATP channels, 5-HD partially inhibits K+ efflux from the ischemic heart (600, 681) and mitigates action potential duration shortening during cardiac ischemia (Table 6). Curiously, 5-HD had no effect on the negative inotropic effects of cromakalim in the absence of ischemia, but completely reversed its cardioprotective effects, which led to the suggestion that 5-HD is an ischemia-selective KATP channel inhibitor (552). Off-target effects of 5-HD are also possible since 5-HD can be activated to 5-HD CoA in mitochondria and metabolized by the β-oxidation pathway, but also blocks β-oxidation of other fatty acids (see sect. VI). Thus, although 5-HD is very effective at mitigating the protective effect of ischemic or pharmacological preconditioning (Table 10), the drug lacks selectivity: published data show that it blocks both mitochondrial and sarcolemmal KATP channels, and it is no easy task to assign a specific molecular target to this compound.
c) genetic mouse models show that sarcolemmal KATP channels are cardioprotective. The basis of this argument is that mice genetically deficient in sarcolemmal KATP channel subunits are less protected against ischemic insults; hence, sarcolemmal KATP channels must be cardioprotective. Indeed, the protective effect of ischemic preconditioning to reduce infarct size or to improve postischemic functional recovery was readily observed in wild-type mice, but was absent in Kir6.2−/− mice (771, 876). As noted above, the preconditioning-like cardioprotective effect of diazoxide is similarly lost in the Kir6.2−/− mice (770, 876). Moreover, the protective effect of late preconditioning, induced by pretreatment with the A3 AR agonist CP-532,903, is absent in the hearts of Kir6.2−/− mice (856). Knockout of Kir6.2 worsens myocardial Ca2+ load in vivo and impairs functional recovery after ischemia (305). It also negates ischemic preconditioning-induced protection of myocardial energetics (306). Consistent with these data, transgenic overexpression of SUR2A in mice generates a cardiac phenotype resistant to ischemia (189). These genetic studies provide powerful arguments for a role of sarcolemmal KATP channels in cardioprotection, but these arguments must be interpreted by keeping in mind two caveats. First, these studies are all performed using mice, which have a high metabolic rate and abbreviated action potentials. Although the shorter action potential of mice per se does not seem to makes a difference with regard to the response to ischemia-induced injury (see above), species differences can be important when considering metabolic effects or other electrophysiological responses (e.g., the development of arrhythmias). Second, these arguments assume that Kir6.2/SUR2A subunits exclusively form the ventricular sarcolemmal KATP channels, which may be a simplistic view, given some reports of their expression in mitochondria (Table 5) and other cardiovascular tissues (see sect. VII).
d) is it the mitochondrial or the sarcolemmal KATP channel that is responsible for cardioprotection? This is probably not the correct question to ask. In all likelihood, sarcolemmal and mitochondrial KATP channels, along with the other KATP channel subtypes in the cardiac conduction system, endothelium, smooth muscle, and even noncardiovascular tissues, are all involved and they synergistically interact to protect the heart against ischemic and other stresses (Figure 13). If so, this might explain the extraordinary efficacy of diazoxide, which affects most of these KATP channels at the concentrations used experimentally to elicit cardioprotection (125). Moreover, although cardiac ischemia is usually characterized as a disease of the myocyte, it is clear that the vasculature, and especially endothelial cells, is a major target of this pathology (412, 627, 716). As is true for cardiac myocytes, endothelial dysfunction during ischemia can be prevented by preconditioning with brief periods of intermittent ischemia (482, 627) in a KATP channel-dependent manner (85, 534), thus extending to coronary endothelial cells the concept of endogenous protection mechanisms. Even though endothelial KATP channel opening mediates the secretion of nitric oxide and ROS (see above), which in turn affect the outcome of ischemic damage (406) and may mediate the protective effects of late preconditioning (778), the role of these channels in cardioprotection and the genesis of preconditioning has largely not been investigated. Likewise, the potential role(s) of smooth muscle KATP channels to regulate blood flow, KATP channels in the specialized conduction system and their contribution to arrhythmogenesis, and atrial KATP channels with their role in regulating ANP release, will all remain largely unexplored until suitable tools become available to investigate these questions.
8. Mechanisms by which KATP channels protect against ischemic damage
a) the energy-sparing hypothesis. KATP channel opening accelerates the time to contractile failure during ischemia, in part due to enhancing action potential shortening, which in turn leads to reduced Ca2+ influx into cells and a negative inotropic effect (Figure 12). The decreased rate of ATP consumption is expected to preserve intracellular nucleotide levels, thus having an anti-ischemic effect. This “ATP-sparing hypothesis” has been postulated in Noma′s patch-clamp study, in which the cardiac KATP channel was initially identified (607). Each of the steps in this protective pathway has now been realized to some extent. Indeed, contractile failure during ischemia is accelerated by KCOs and mitigated by KATP channel blockers (556). Moreover, KATP channel activation during ischemia mitigates Ca2+ overload (142, 480, 764), preserves mitochondrial integrity (764) and leads to energy preservation (556) (see below).
b) protection of mitochondrial integrity. Preconditioning and KATP channel opening slows energy metabolism and protects mitochondrial integrity during ischemia (196, 245, 264, 796, 887). This protection may potentially occur directly as a result of the activity of mitochondrial KATP channels (611). This idea is that the K+ uptake into mitochondria during ischemia, together with the activity of a K+/H+ antiporter, better maintains matrix volume regulation, which in turn stimulates the rate of oxidative phosphorylation (257) (Figure 12). Although this paradigm remains to be proven in the setting of the ischemic heart, mitochondria isolated from preconditioned hearts do have an elevated rate of ATP synthesis (245). Curiously, in the latter study, although 5-HD partially reversed the ATP synthesis rate, diazoxide or HMR-1098 (when applied before ischemia) did not stimulate the rate of ATP synthesis. KATP channel openers such as diazoxide may have additional energetic benefits. For example, in perfused rat hearts, diazoxide was found to reduce cellular and mitochondrial ATPase activities, along with nucleotide degradation, which may contribute myocardial ATP preservation during ischemia (196). Furthermore, proteomic analysis of diazoxide “preconditioned” rabbit ventricular myocytes showed a remarkable plasticity in the expression and phosphorylation profiles of proteins involved in mitochondrial energetics, including subunits of tricarboxylic acid cycle enzymes and oxidative phosphorylation complexes (20). The effects of these compounds, however, may be complex since KATP channel openers such as diazoxide and pinacidil have opposite effects on mitochondrial respiration under different energetic conditions (669). Mitochondrial integrity may also potentially be indirect in that sarcolemmal KATP channel opening diminishes intracellular Ca2+ overload, which in turn may protect mitochondrial function (764). KATP channel opening in the vasculature may also contribute to increase blood flow to the ischemic tissue and thus to lessen the damaging effects of ischemia.
c) other mechanisms. During ischemia, reperfusion, and ischemic preconditioning, a variety of mechanisms may affect KATP channel function, and thereby directly or indirectly contribute to the role of KATP channels during these events. Receptor and molecular signaling represents just one of these and will not be discussed in detail since an excellent recent review exists on this subject (806). In brief, the KATP channel can be regulated by several signaling cascades, including cAMP-mediated events. The effects of cAMP are either mediated by PKA (59, 625, 728) or through a PKA-independent pathway that is mediated by EPAC proteins (420). Although cAMP has been studied in the setting of the ischemic heart for many years (71, 201), the role of these signaling cascades on KATP channel in the ischemic setting is not well defined. Another signaling molecule, PKC, has a prominent role by initiating protection in various forms of IPC (514). Endogenous mediators (autacoids) of IPC, such as adenosine, signals through PKC signaling pathways (279). The mechanisms by which PKC confers cardioprotection are elusive and may involve preservation of mitochondrial function, ROS generation, and improved cardiac overload (195, 895). Both mitochondrial and sarcolemmal KATP channels have been proposed as effectors of PKC signaling (320, 436, 499, 503, 515, 752, 826, 861). The PKC family consists of 10 members, grouped into classical (c), novel (n), and atypical classes (759). The Ca2+-independent nPKCε isozyme (217) has a well-described protective role in ischemia/reperfusion and ischemic preconditioning (369, 506, 759). The role of nPKCδ is controversial, since both protective and detrimental roles have been described (370, 551, 759). Interestingly, an increase in intracellular Ca2+ by itself can mediate ischemic preconditioning (568), and there is strong evidence that the protective effect is mediated by activation of the Ca2+-dependent cPKCα isozyme (78, 333, 440, 617, 817, 893, 908, 923). KATP channels are positively regulated by molecular signaling pathways relevant to IPC. Patch-clamp data demonstrate that PKC activates cardiac KATP channels directly (6, 364, 365, 384, 501, 823, 937) and upregulates KATP channel surface density (199, 824). Moreover, the Ca2+-dependent and Ca2+-independent PKCs may act synergistically to activate KATP channels (36, 384). It has also been shown that specific PKCε activator (ψεRACK) peptides (185) prime KATP channels to open during simulated preconditioning in isolated cardiomyocytes. In smooth muscle, however, Ca2+-mediated PKC activation may cause internalization of the KATP channels (36). A possible role for PKCδ (in concert with AMPK and p38) in regulating the KATP channel surface density in ventricular myocytes has been suggested (824). Apart from these studies, the effects of different PKC isozymes on KATP channel function and trafficking during IPC have not been fully studied.
The downstream effectors of the PKC pathway remain to be fully identified and include AMPK (911), which has a demonstrated cardioprotective role in ischemia/reperfusion and ischemic preconditioning (98, 101, 112, 367, 467, 751, 885, 912). AMPK may signal through multiple mechanisms, including an altered glucose metabolism (520) and enhanced glucose uptake during ischemia (398, 677), which is mediated by the translocation of the glucose transporter GLUT4 from endosomes to the sarcolemma (471, 493, 604) and a reduction in GLUT4 endocytosis (894). A role for KATP channels was demonstrated by the finding that Kir6.2 knockout reversed the cardioprotection elicited by AMPK activation (190). It appears that AMPK can affect KATP channels in at least two ways: AMPK activation increases the open probability of ventricular KATP channels (905), but also stimulates translocation of KATP channels to the surface membrane; at least in pancreatic β-cells (109, 504, 632) (also see sect. VI). Studies with cellular models of ischemic or pharmacological preconditioning have suggested that KATP channel translocation to the surface might also be an important element of AMPK signaling in the ischemic heart (767, 824). Although a functional role for endosomal KATP channels has been suggested (47), trafficking of KATP channel as a protective mechanism during ischemic preconditioning in intact hearts has not been directly investigated.
A myriad of other intracellular events and signaling cascades may potentially affect KATP channel synthesis, trafficking, function, and degradation during ischemia and ischemic preconditioning. A few of these have started to emerge in recent publications. For example, cGMP, which has a known protective role in ischemic preconditioning (517), has been shown to stimulate cardiac KATP channels via a ROS/calmodulin/CaMKII signaling cascade (103). Other studies, however, suggested a negative role for CaMKII activation in the regulation of KATP channel density by phosphorylation of specific tyrosine residues in the Kir6.2 COOH terminus and potentially influencing its interaction with the μ2 subunit of the AP2 adaptor complex, and thus promoting internalization (495, 741). Much research is needed to elucidate the relevance of these findings in KATP channel function and trafficking during cardiac ischemia and IPC.
9. KATP channels in the coronary vasculature
Smooth muscle KATP channels have a widely acknowledged role in maintaining basal coronary vascular tone in response to various physiological stimuli (see above). They also have an established role for the increased blood coronary flow in response to hypoxic conditions. The involvement of KATP channels was first revealed when it was found that glibenclamide blocks hypoxia-induced vasodilation in coronary arteries (161). Thus stress events that impair energy metabolism and intracellular nucleotide concentrations appear to be sufficient to activate KATP channels in coronary arterioles, which in turn causes membrane hyperpolarization decreased Ca2+ fluxes and vasodilation (83).
a) coronary blood flow during ischemia. Myocardial ischemia most often originates from impaired blood flow. Even in vessels not directly affected (e.g., by a blood clot), myocardial ischemia can severely compromise coronary vasodilation and blood flow, which has direct negative consequences for the fate of cardiac myocytes during ischemia/reperfusion. It is therefore important to consider the changes that ischemia may have on blood flow regulation. The evidence for an involvement of KATP channels in postischemic coronary blood flow is scant. Ischemia has been shown to induce rapid alterations in local myocardial blood flow patterns (572), which impacts local perfusion. Myocardial hypoxia was suggested to be the major factor stimulating microvascular vasodilation of resistance vessels distal to the coronary artery stenosis (702), and it is likely that hypoxia signals via KATP channel opening in the ischemic setting.
b) reactive hyperemia. In many organs, including the heart, a brief period of ischemia is inevitably followed by a large and transient increase in blood flow. This phenomenon, called reactive hyperemia, is used as an index of coronary flow reserve (129). KATP channels appear to have a key role in the mechanism of coronary reactive hyperemia since glibenclamide inhibits the peak blood flow responses and the duration of the hyperemic period in mouse, dog, guinea pig, and human hearts (33, 46, 121, 192, 913). KATP channels in both the coronary microvasculature and larger vessels are involved in this response (453). Coronary microvascular dysfunction results in a reduction of the flow reserve and a failure of myocardial oxygen consumption to meet demand. Although a role for KATP channels in the coronary smooth muscle cells is easily postulated, it appears that the endothelium may also be involved since endothelium removal reduces the vasodilatory response to KATP channel openers and the ischemic reactive hyperemic response in coronary arteries in the instrumented dog heart (472). More work is needed to investigate the contribution of different KATP channels in the vascular bed and how (if at all) they relate to other mechanisms of reactive hyperemia, such as nitric oxide production (453).
c) the “no reflow” phenomenon. A therapeutic goal during reperfusion of previously ischemic tissue (e.g., following angioplasty) is to restore blood flow as rapidly and effectively as possible. However, as initially described by Kloner et al. (451), successful restoration of coronary flow does not necessarily translate into improved tissue perfusion. This “no reflow” phenomenon, which is responsible for poor clinical prognosis (429, 430), is characterized by microvascular and endothelial dysfunction (561). Several studies have implicated a role for KATP channels in the no-reflow phenomenon. In a mini-swine model of regional ischemia, simvastatin was found to limit the area of no-reflow after ischemia/reperfusion, determined by myocardial contrast echocardiography and histological evaluation (896). In this study, postinfarction treatment with glibenclamide abrogated the protective effect of simvastatin, suggesting that simvastatin reduced myocardial no-reflow by KATP channel activation. In another mini-swine study, remote periconditioning (brief cycles of ischemia and reperfusion of a remote organ applied during sustained myocardial ischemia) decreased the area of no-reflow in a glibenclamide-sensitive manner, also suggesting an involvement of KATP channels (926). No-reflow is also reduced by KATP channel openers, angiotensin-converting enzyme inhibitors, angiotensin receptor blockers, beta receptor blockers and adenosine, and in each case the protection is lost by coadministration of glibenclamide (924, 925, 927, 928). It is possible that at least some of the protective effect may be mediated by modulating the production or release of vasoactive substances (924, 925, 927). This is a potentially important and therapeutically relevant field of study and further mechanistic studies are warranted.
C. Hypertension and Dilated Cardiomyopathy
Congestive heart failure occurs when a mismatch exists between the ability of the heart to deliver blood and the needs of the body. Common causes of heart failure include cardiac insufficiency due (among others) to myocardial infarction and high blood pressure, which in turn are caused by coronary atherosclerosis and kidney disorders. Although not a direct cause of heart failure, experimental evidence suggests that KATP channels may be involved. For example, in a dog heart failure model with ventricular pacing, an elevation in left atrial pressure was described. Concomitant with the atrial stretch, plasma atrial natriuretic peptide (ANP) levels and urinary sodium excretion were elevated (107). The release of ANP from atria is known to be modulated by KATP channel opening (see sect. VIIB). Consistently, it was found that the KATP channel blocker glibenclamide delayed the increase in plasma ANP levels and decreased urinary Na+ excretion. These data point to a possible protective role for atrial KATP channels in the compensatory release of atrial ANP in response to an elevated blood pressure. The Kir6.2−/− mice have clarified a further role for KATP channels: in an experimental hypertension model consisting of a combination of a left nephrectomy, long-term treatment with deoxycorticosterone and a high-salt diet, the knockout mice exhibited a decreased survival rate and these mice more readily developed ventricular remodeling and heart failure (419), associated with unfavorable proteomic remodeling as determined by mass spectrometry (943). Moreover, with pressure overload imposed on the left ventricle by transverse aortic constriction, Kir6.2−/− mice displayed compromised myocardial performance with elevated left ventricular end-diastolic pressure followed by biventricular congestive heart failure. Glibenclamide could recapitulate this phenotype in wild-type mice (890). Proteomic profiling of KATP channel-deficient hypertensive heart maps risk for maladaptive cardiomyopathic outcome. Combined, these data suggest that KATP channels have a protective role in the transition from conditions of imposed hemodynamic load to the development of maladaptive pathological remodeling and heart failure. The relationship between KATP channels and the development of failure are not clear at present.
X. GENETIC VARIATION IN KATP CHANNEL GENES: RELEVANCE TO CARDIOVASCULAR DISEASE
Genetic variation and mutations in KATP channel subunit genes have been linked to the etiology of a number of life-threatening human diseases (136, 599, 620). A number of mutations in KCNJ11 (Kir6.2) and ABCC8 (SUR1) have been associated with insulin disorders, including permanent neonatal diabetes (OMIM: 606176), transient neonatal diabetes (OMIM: 610582), familial hyperinsulinemic hypoglycemia (OMIM: 601820), and susceptibility to type 2 diabetes mellitus (OMIM: 125853). These will not be discussed, and the reader is referred to reviews on this subject (238, 592). Mutations in these two genes are generally not associated with cardiovascular disease. An exception is a polymorphism that results in an E23K amino acid substitution in the Kir6.2 NH2 terminus. Homozygosity for K23 (the KK variant), which occurs more frequently in type II diabetic than in control subjects (181, 319, 715), is also overrepresented in individuals with dilated cardiomyopathy and congestive heart failure compared with controls, and is also associated with abnormal exercise stress testing (664). The K23 KK genotype was additionally associated with greater left ventricular size among subjects with increased stress load due to hypertension, suggesting that the Kir6.2 K23 polymorphism may be a risk factor for adverse subclinical myocardial remodeling (665).
Recent data demonstrate associations in genetic variation in KCNJ8 (Kir6.1) and ABCC9 (SUR2) with cardiovascular disease (Table 11). For example, genetic variation in KCNJ8 has been linked to risk of Brugada syndrome, J wave syndromes, ventricular fibrillation, atrial fibrillation, and early infantile epileptic encephalopathy. Interestingly, it is the same genetic variant, a c.1265C-T transition that results in a S422L substitution in the Kir6.1 COOH terminus, which is linked to all of these conditions. While clearly important, it is not obvious how this variant has such diverse phenotypes, especially given that Kir6.1 is not generally thought of as a subunit that gives rise to the ventricular KATP channel (Table 2). Moreover, although the variant is present at low frequency in European Americans and African-American individuals, it is present in up to 4% of Ashkenazi Jews (838). Patch-clamp studies report that the S422L mutant Kir6.1 channels have a gain of function due to a reduced ATP sensitivity, thus resulting in incomplete closing of the channel under normoxic conditions (49, 557). Recent data also demonstrate a link between KCNJ8 genetic variation and hypertrichotic osteochondrodysplasia, also named Cantú syndrome (281, 599), in which hypertrichosis leads to thick scalp hair and other abnormalities. Cardiac and vascular abnormalities occur in ∼80% of these cases and may include patent ductus arteriosus, ventricular hypertrophy, pulmonary hypertension, pericardial effusions and tortuous retinal vessels, and multiple tortuous pulmonary arteriovenous communications (281). Two KCNJ8 variants, resulting in Kir6.1 amino acid substitutions V65M or C176S, have been linked to this disorder. Several ABCC9 variants have also been linked to Cantú syndrome, resulting in SUR2 amino acid substitutions including H60Y, A478V, S1020P, C1043Y, R1116H, R1116C, R1154W, and R1154Q (Table 11). SUR2 mutations (A1513T, early truncation at L1524 and T1547I) have also been linked to dilated cardiomyopathy and familial atrial fibrillation.
It has been known for a long time that KATP channels are expressed ubiquitously and that they have diverse functions in different cell types (24). In some cell types, such as pancreatic β-cells, the physiological role(s) of KATP channels are characterized better than others. Molecular cloning of subunits of plasma/sarcolemmal KATP channels, the development of transgenic and knockout mice targeting these subunits, and the relationship between human disease and genetic variations in the genes coding for these subunits are instrumental in reaching a more complete understanding of the rich diversity of KATP channels within components of the cardiovascular system (SA node, atria, specialized conduction system, ventricle, smooth muscle, and endothelium). Pharmacological approaches have been similarly helpful in underscoring the important roles of KATP channels in vascular and cardiac function in cardiovascular pathologies, but these approaches are limited due to the overlapping pharmacological profiles of most of these KATP channel modulating compounds and the fact that the effects of many of these compounds are altered by the cellular metabolic state. Molecular identification of each of the KATP channel types, particularly the mitochondrial KATP channel, and the development of small molecules with better target selectivity, will be of enormous benefit in moving forward the field. Undue emphasis on the functional and pathophysiological role of any one of the KATP channels, however, should be avoided since it is likely that the various types of KATP channels in the different cardiovascular tissues synergistically interact to protect against stress. It is even likely that KATP channels in noncardiovascular tissues, such as cardiac neurons, fibroblasts, and macrophages, may help to shape the cardiovascular responses to stress by regulating the release of hormones (ACh and NE) and being involved in inflammatory responses. It is hoped that future studies may find a way to comprehensively investigate the diverse roles of KATP channels in the cardiovascular system.
This work was supported by National Institutes of Health Grants HL085820, HL119209, and HL126905 and in part by the New York Masonic Seventh District Association, Inc.
No conflicts of interest, financial or otherwise, are declared by the authors.
Address for reprint requests and other correspondence: W. A. Coetzee, NYU School of Medicine, Alexandria Center for Life Science 824, 450 East 29th St., New York, NY 10016 (e-mail:).
- Copyright © 2016 the American Physiological Society