The Slo(w) path to identifying the mitochondrial channels responsible for ischemic protection

Abstract

Mitochondria play an important role in tissue ischemia and reperfusion (IR) injury, with energetic failure and the opening of the mitochondrial permeability transition pore being the major causes of IR-induced cell death. Thus, mitochondria are an appropriate focus for strategies to protect against IR injury. Two widely studied paradigms of IR protection, particularly in the field of cardiac IR, are ischemic preconditioning (IPC) and volatile anesthetic preconditioning (APC). While the molecular mechanisms recruited by these protective paradigms are not fully elucidated, a commonality is the involvement of mitochondrial K⁺ channel opening. In the case of IPC, research has focused on a mitochondrial ATP-sensitive K⁺ channel (mitoK_ATP), but, despite recent progress, the molecular identity of this channel remains a subject of contention. In the case of APC, early research suggested the existence of a mitochondrial large-conductance K⁺ (BK, big conductance of potassium) channel encoded by the Kcnma1 gene, although more recent work has shown that the channel that underlies APC is in fact encoded by Kcnt2. In this review, we discuss both the pharmacologic and genetic evidence for the existence and identity of mitochondrial K⁺ channels, and the role of these channels both in IR protection and in regulating normal mitochondrial function.

Ischemia–reperfusion injury and protection

Ischemia, defined as the blockage of delivery of oxygen and nutrients to tissues, is a pathologic event that underlies some of the most prevalent causes of death in humans. Paradoxically reperfusion (i.e., the re-establishment of oxygen and nutrient delivery) is also a pathologic event. Taken together, these events comprise ischemia–reperfusion (IR) injury, the underlying cause of diverse conditions such as heart attack and stroke. The focus of this review is cardiac IR; in the US alone, there are 750 000 heart attacks a year, killing 116 000 people. In addition, over 300 000 patients undergo a ‘scheduled’ cardiac ischemic event when the heart is arrested and placed on bypass during open heart surgery [1]. Since cardiac IR injury is a major cause of mortality and morbidity, it is surprising that beyond reperfusion itself (e.g., thrombolysis or balloon angioplasty), there are virtually no drug therapies to acutely treat it [2,3].

The heart is an energetically demanding tissue, with the bulk of its ATP demand met by mitochondrial oxidative phosphorylation [4,5]. Upon ischemia, mitochondrial ATP synthesis halts, starving processes such as actin/myosin cross-bridge cycling and the maintenance of ion gradients by the Na⁺/K⁺-ATPase and sarco/endoplasmic reticulum Ca²⁺-ATPase (SERCA). In addition, glycolytic metabolism generates lactate, causing cellular acidosis which then activates the Na⁺/H⁺ exchanger and leads to a rise in intracellular Na⁺ [6,7]. Na⁺ export is driven by the Na⁺/Ca²⁺ exchanger, leading to a rise in cytosolic Ca²⁺ [8], which is compounded by the ATP-starved SERCA pump [9,10]. A cytosolic Ca²⁺ overload ensues, with some Ca²⁺ entering the mitochondrion. However, these events alone are insufficient to trigger opening of the mitochondrial permeability transition (PT) pore, since acidic pH and a reduced pyridine nucleotide pool (NADH) maintain the PT pore in a closed state [11–13]. At reperfusion, further Ca²⁺ overload occurs [14,15], pH rebounds [16], and a burst of reactive oxygen species (ROS) generation occurs as metabolites accumulated during ischemia are rapidly oxidized [17]. This combination of Ca²⁺, pH and ROS triggers opening of the PT pore, leading to cell death [18–25]. These events are summarized in Figure 1.

Figure 1. Schematic representation of pathologic events during ischemia and reperfusion.

Key events are listed below each figure and described in detail in the text (Section ‘Ischemia-–Reperfusion Injury and Protection’). PMCA, plasma membrane Ca²⁺-ATPase; NCX, Na⁺/Ca²⁺ exchanger; NHE, Na⁺/H⁺ exchanger; FA, fatty acids; PT, permeability transition pore.

Given the universality of IR injury as a pathologic insult in biology, it should not be surprising that a diverse array of organisms [26–31] exhibit mechanisms to limit damage due to this insult. Among the best studied of such mechanisms is ischemic preconditioning (IPC), in which short periods of prior IR afford protection against subsequent IR injury. IPC is an example of hormesis (i.e., ‘what doesn’t kill you makes you stronger’) and is clinically applicable in humans [32]. IPC affords protection in two phases: the first develops in minutes, lasts 2–3 h, and involves cell signaling cascades that terminate at mitochondria [4], as will be discussed here. A second protective phase develops in ~24 h and lasts up to 72 h, requiring gene transcription and de novo protein synthesis [33], but will not be considered further. Of particular interest for this review, it is also known that halogenated volatile anesthetics (halothane, isoflurane, sevoflurane, and desflurane) can mimic the protection afforded by IPC, a process known as anesthetic preconditioning (APC) [34–36].

The centrality of mitochondria to IR pathology has driven the organelle to be a natural focus for research on IR protection. In this regard, a common mechanism believed to underlie several cardioprotective paradigms, including IPC and APC, is the opening of potassium channels in the mitochondrial inner membrane [37–42]. This review will focus on the evidence for existence and identity of these channels.

Mitochondrial K⁺ homeostasis and discovery of mitochondrial K⁺ channels

A note on nomenclature

Before discussing these channels in detail, nomenclature should be clarified. The gene encoding the mitochondrial ATP-sensitive K⁺ (K_ATP) channel is unknown, and so here we use the term ‘mitoK_ATP’. For other K⁺ channels, where possible the International Union of Basic and Clinical Pharmacology (IUPHAR) nomenclature is used [43] (see the Abbreviations list); however, there are many alternative names commonly found in the literature, described here.

The term ‘BK’ was coined in 1984 when a ‘big K⁺’ or large-conductance K⁺ channel activated by Ca²⁺ was recorded by patch clamp [44]. In 1986, the Drosophila slowpoke (Slo) mutation was shown to abolish a Ca²⁺-activated K⁺ current [45], and subsequently, the ‘Slo1’ gene was shown to be conserved among phyla. Hence, BK (also known as ‘maxi-K’) was the term used to describe the channel encoded by the Slo1 gene. This gene is now known as Kcnma1, and the channel is known as as K_Ca1.1.

A related family of Slo genes has since been identified, which now includes Slo1 (Kcnma1), Slo2.1 (Kcnt2), Slo2.2 (Kcnt1), and Slo3 (Kcnu) [46–52]. Kcnma1 and Kcnu encode Ca²⁺-activated K⁺ (K_Ca) channels (now known as K_Ca1.1 and K_Ca5.1, respectively). Lower organisms such as Caenorhabditis elegans have a single gene (termed ‘Slo2’) encoding a K_Ca channel [53], and here, we use the name SLO2 to refer to this channel in C. elegans. In contrast, this gene has diverged into two paralogs in mammals, and somewhat confusingly, the mammalian channels are, in fact, Na⁺-activated (K_Na) channels: the gene previously known as Slo2.1 was thought to encode a channel termed K_Ca4.2 (also known as ‘Slick’). This gene is now known as Kcnt2 and encodes a channel known as K_Na1.2. The gene previously known as Slo2.2 was thought to encode a channel termed K_Ca4.1 (also known as ‘Slack’). This gene is now known as Kcnt1 and encodes a channel known as K_Na1.1. The umbrella term ‘K_Na1.x’ is used here to refer to both mammalian K_Na1.2 and K_Na1.1 channels. Additional naming complexity is also imparted due to the channels encoded by the Slo family genes having alternate splice variants that can heteromultimerize [46,47,50,54,55] (see Sections ‘Mitochondrial K_Ca1.1 Channels and Mitochondrial K_Na1.x Channels’).

The general term ‘K_Ca’ refers to the small-conductance (SK) channels, the intermediate-conductance (IK) channels [56], and the channel encoded by Slo1 (Kcnma1). Unfortunately, ‘K_Ca’ is also sometimes used to reference all channels encoded by the Slo gene family, even though it is now apparent that many of these are K_Na channels (see above).

In the mitochondrial field, the terms ‘mitoBK’ and ‘mitoK_Ca’ (and sometimes even ‘mitoBK_Ca’!) have been used interchangeably. Wherever possible, we attempt to define these channels using the IUPHAR nomenclature. However, many studies have assigned channel names on the basis of pharmacology alone, prior to the advent of molecular biological identification. Therefore, in such cases, we default to the nomenclature system used by the authors of these studies.

Mitochondrial K⁺ channels

The mitochondrial inner membrane, while maintaining a tight barrier to proton permeability that is essential for its bioenergetic function, is also selectively permeable to numerous cations (K⁺, Ca²⁺, Mg²⁺, and Na⁺) [57–59] and anions (Cl⁻, PO₄⁻, nucleotide phosphates, and di- and tri-carboxylates) [60]. This selective permeability is under the control of membrane transporters, and here we focus on those ion channels mechanistically linked to IR protection — namely, the mitoK_ATP channel and the K⁺ channels encoded by the Slo gene family. Other mitochondrial K⁺ transport proteins [e.g., the K⁺/H⁺ exchanger (KHE) and voltage-gated K⁺ channels] are reviewed extensively elsewhere [61–63].

Mitochondrial K⁺ permeability has been studied since the early days of bioenergetics [64–66]. K⁺ entry into mitochondria is accompanied by osmotically obliged water, resulting in swelling and a decreased refractive index [67], making mitochondrial volume (easily measured spectrophotometrically as the scattering of light by isolated mitochondrial suspensions) a useful surrogate measure of K⁺ uptake [68–70]. Energetically driven swelling in K⁺-containing buffers was initially attributed to an inherent K⁺ permeability of the mitochondrial membrane [71,72], and subsequent studies identified an electrically neutral KHE supporting the existence of a mitochondrial K⁺ cycle [73] (Figure 2). The fact that both K⁺ influx and efflux consume the transmembrane H⁺ gradient suggests the functional importance of the cycle, and it has been suggested that the cycle serves to regulate mitochondrial volume [74], which in turn may be an important regulator of respiratory function [75,76]. Alternatively, it has been proposed that mitochondrial K⁺ homeostasis serves to regulate ROS production [77,78].

Figure 2. Mitochondrial K⁺ cycle.

The mitochondrial inner membrane potential (Δψ_m), which is generated by the respiratory chain (complexes I–IV [396,397], left), drives K⁺ entry into the mitochondrial matrix through either small- or intermediate-conductance channels (e.g., K_ATP, K_Ca2.x, or K_Ca3.1) or large-conductance channels (K_Na1.2 or K_Ca1.1). This K⁺ current is followed by osmotically obliged water, resulting in swelling of the matrix. K⁺ is removed from the matrix through the KHE that also consumes Δψ_m. The outer membrane is largely permeant to all solutes and hence is depicted as a dotted line.

The first report of a bona fide mitochondrial K⁺ channel by patch-clamp electrophysiology was in 1991 [79] and opened the way for identification of K⁺ currents attributable to known K⁺ channel families based on electrophysiological properties. Numerous K⁺ channels have now been reported in mitochondria, including K_V1.3 in lymphocyte mitochondria [80], a K_Ca channel in liver [81] and fibroblast [82] mitochondria, and K⁺ATP in mitochondria from glioma [79] and cardiac ventricles [83]. Additional methods supporting mitochondrial K⁺ channel identity include the following: (i) immunologic detection such as western blot [84] and fluorescent immunocytochemistry [85]. (ii) Mitochondrial fractionation and reconstitution of channels into liposomes [86]. (iii) Indirect measurement of mitochondrial K⁺ uptake by fluorescent probes such as potassium binding fluorescent indicator [87] or fluorescent measurement of mitochondrial Tl⁺ uptake as a surrogate for K⁺ flux [68]. (iv) Genetic tagging of candidate K⁺ channel proteins and their tracking to mitochondria within cells [88]. (v) Sensitivity of these measurements to a variety of pharmacologic agents that are known to act on particular classes of K⁺ channel (see Sections ‘Mitochondrial K_ATP Channel: Composition, Pharmacology, Regulation, Role in IR Protection; Mitochondrial K_Ca2.x and K_Ca3.1 Channels: Composition, Pharmacology, Regulation, Role in IR Protection; and Mitochondrial K_Ca1.1 Channels and Mitochondrial K_Na1.x Channels’). (vi) Generation of mice or cell lines with candidate mitochondrial K⁺ channel genes deleted [89–91].

While these considerable efforts support the existence of bona fide K⁺ channels in mitochondria, their molecular identities are still hotly debated. In particular, the mitoK_ATP channel is controversial [92–95]. This topic has been extensively reviewed elsewhere and so will be discussed only briefly here in the Section ‘Mitochondrial K_ATP Channel: Composition, Pharmacology, Regulation, Role in IR Protection’. The identity of a mitochondrial large-conductance K⁺ (BK) channel is also unclear [42,96–98] and is discussed in detail in Sections ‘Mitochondrial K_Ca2.x and K_Ca3.1 Channels: Composition, Pharmacology, Regulation, Role in IR Protection and Mitochondrial K_Ca1.1 Channels and Mitochondrial K_Na1.x Channels’. For reference, a schematic of selected cardioprotective stimuli and their proposed target mitochondrial K⁺ channels, along with inhibitors of such cardioprotection, is shown in Figure 3.

Figure 3. Cardioprotective stimuli, molecular targets, and inhibitors.

A subset of pharmacophores in Table 1 is known to confer cardioprotection. These are shown at the top of this figure, along with the cardioprotective stimuli of APC and IPC. Target mitochondrial K⁺ channels are depicted below (blue), with known pharmacologic inhibitors of these protective paradigms in red. As detailed in the text, a combination of genetic and pharmacologic information has demonstrated that K_Ca1.1 and K_Na1.2 each participate in distinctly activated mechanisms of cardioprotection.

Mitochondrial K_ATP channel: composition, pharmacology, regulation, and role in IR protection

A study of the mitoK_ATP channel is incontrovertibly linked to the study of IR protection; much of the evidence for the existence of the channel comes from effects of channel-modulating drugs on cardiac IR injury, and much of the evidence for cardioprotection comes from the design and application of agents targeting putative mitoK_ATP channels. Thus, research into this channel’s composition or pharmacology is largely driven by its role in cardioprotection [99–101].

ATP-sensitive K⁺ channels (K_ATP) have been detected in numerous cell membranes including plasma membrane [102], sarco/endoplasmic reticulum [103], mitochondria [79], and the nuclear envelope [104]. Generally, K_ATP channels are octamers composed of four 2-transmembrane inward-rectifying K⁺ channel (KIR) subunits (KIR6.1/6.2), plus four 17-transmembrane sulfonylurea receptor (SUR) subunits (SUR1/2A/2B) [105]. The ventricular myocyte surface K_ATP channel comprises KIR6.2/SUR2A [106] and regulates both cell volume and action potential duration [107].

Initial observations suggested a role for surface K_ATP channels in IPC [108–110], with depressed contractility thought to be the mechanism of cardioprotection [107,111,112]. However, the discovery that K_ATP activators [diazoxide (DZX) [113], cromakalim [114], and aprikalim [115]] were capable of protecting noncontracting myocytes, in a manner blocked by K_ATP inhibitors [5-hydroxydecanoate (5-HD), glyburide, and HMR1098 [116–118]], suggested a protective mechanism independent of depressed contractility. The discovery of a mitochondrial K_ATP channel with sensitivity to DZX and cromakalim [79,83,119] provided a candidate mechanism, with further support provided by evidence that cardiac surface K_ATP channels are insensitive to DZX [120–123]. Subsequently, a channel with pharmacologic properties ascribed to mitoK_ATP was recognized as a key player in IPC signaling, despite an ongoing debate regarding the molecular identity of this channel. Weak evidence also exists for a potential role of a mitochondrial K_Ca channel in IPC [124,125] and will be discussed in Section ‘Mitochondrial K_Ca1.1 Channels and Mitochondrial K_Na1.x Channels’.

Although global knockout mice exist for K_ATP channel subunits (i.e., KIR6.1/6.2 and SUR1/2), several complications preclude their use to study IPC. For example, Kir6.1^−/− mice [126] exhibit a form of angina, and it is known that patients with unstable angina exhibit a preconditioned phenotype [127,128]. While Kir6.2^−/− mice [91,121] exhibit blunting of protection by IPC, these mice also have impaired insulin secretion and mild glucose intolerance [91], and it is known that diabetes abrogates protection by IPC [129]. Similarly, both Sur1^−/− and Sur2^−/− mice exhibit glycemic disturbances and are endogenously protected against cardiac IR injury [130–132], precluding their use to study cardioprotective signaling. Furthermore, the Sur1^−/− mouse was demonstrated to still express alternate splice variants [133], again confounding studies attempting to assign functions to the SUR1 protein. Owing to these confounds, assignment of a particular combination of KIR/SUR subunits as underlying IPC has not been possible to date. In addition, evidence favoring a mitoK_ATP composition of the canonical KIR6/SUR proteins needs to be balanced against the discovery that the original antibodies used to identify these proteins in purified mitochondria recognize off-target proteins unrelated to K⁺ channel function [134]. Furthermore, although a smaller 55 kDa splice variant of SUR2A has been reported in mitochondria [135], this was detected using custom antibodies which failed to detect the same 55 kDa band in similar samples from the same laboratory a year earlier [133]. A subsequent study [136] also suggested rather equivocal evidence for the existence of this 55 kDa band, and overall caution should be used in interpreting any immunologic evidence for a mitoK_ATP channel.

The pharmacology of mitoK_ATP is conserved in humans [137], rats [79], plants [138], amoeba [139], trypanosomes [140], and C. elegans [141], and remains the default method to assign a role for mitoK_ATP in IR protection. A catalog of K⁺ channel pharmacophores applicable to mitochondrial research is given in Table 1. While many channel modulators are available, the two most commonly linked to mitoK_ATP are the channel-activating benzothiadiazine derivative DZX and the antagonist 5-HD [142]. We see that 10 μM DZX is a relatively specific mitoK_ATP agonist and mimics the protective effects of IPC [113,116], whereas 5-HD prevents both IPC- and DZX-mediated IR protection [143,144]. At concentrations >40 μM, DZX has several other mitochondrial effects (e.g., complex II inhibition and protonophoric activity [145,146]). The specificity of 5-HD has also been questioned, since it undergoes β-oxidation to yield 5-HD-CoA and other derivatives [147,148]. However, this compound is also effective within 1 s in mitoK_ATP assays, suggesting that such metabolism is irrelevant for its acute effects on the mitoK_ATP [149]. Additional K_ATP channel activators including cromakalim and pinacidil [150–153] (Table 1) have also been studied in the context of cardiac IR protection and are thought to elicit protective effects via the mitoK_ATP.

Table 1.

Commonly used pharmacologic agents in the fields of mitochondrial K⁺ channel research and cardioprotection, for K_ATP, K_Ca2.x, K_Ca1.1, and K_Na1.x channels

Channel	Actions	Drugs	EC50/IC50	Refs
KATP	Activators	Atpenin A5	10 nM	[39]
		Cromakalim	1 μM	[113,114,398]
		Diazoxide	10 μM	[113]
		Pinacidil	50 μM	[115,399,400]
	Inhibitors	Fluoxetine	2.4 μm	[68]
		Glyburide	50 μm	[113,119]
		5-HD	100 μM	[116]
		Quinine	100 μM	[119]
		HMR1098	100 μM	[119]
KCa2.x	Activator	DECBIO	3 μm	[180]
	Inhibitors	ChTx	50 nM	[191]
		Apamin	1 μM	[191]
		Fluoxetine	9 μM	[192]
		NS8593	10 μM	[180]
KCa1.1	Activators	Emodepside	14 nM	[258]
		Rottlerin	500 nM	[238]
		NS11021	500 nM	[242]
		NS004	10 μM	[401]
		NS1619	10 μM	[242]
		17-β Estradiol	30 μM	[313]
		Niflumic acid	33 μM	[309]
		Ethanol	20 mM	[315]
	Inhibitors	SloTx	1.5 nm	[402]
		IbTx	50 nM	[232]
		Charybdotoxin	200 nM	[403]
		Paxilline	1 μm	[42]
KNa1.x	Activators	Niclosamide	2.9 μM	[317]
		loxapine	4.4 μM	[317]
		17-β Estradiol	10 μM	[314]
		Bithionol	10 μM	[310]
		Isoflurane	300 μM	[307]
		Niflumic acid	2.1 mM	[294]
	Inhibitors	Bepridil	500 nM	[310]
		Pax	1 μM	[97]
		Verapimil	100 μM	[291]
		Quinidine	100 μM	[308]
		Clofilium	109–331 μM	[290]

A common feature that has arisen in the field of mitoK_ATP pharmacology is complex II of the mitochondrial respiratory chain [39,145,149,154–157]. In short, several compounds that open the mitoK_ATP channel are known to be complex II inhibitors, and in turn many complex II inhibitors have been discovered to open the channel. Among these, the most potent is the complex II inhibitor atpenin A5 (AA5), which is an effective mitoK_ATP activator at low nM concentrations and is cardioprotective in a manner blocked by 5-HD [39,157,158]. Additionally, a mitochondrial membrane fraction enriched in complex II, mitochondrial ATP-binding cassette protein 1, phosphate carrier, adenine nucleotide translocator, and ATP synthase was shown to have mitoK_ATP-like activity when reconstituted in lipid bilayers [86]. The exact nature of the relationship between complex II and the mitoK_ATP is reviewed extensively elsewhere [159]. Similarly, an interaction between complex IV and a paxilline-sensitive K⁺ channel has been reported in membranes isolated from brain mitochondria [160], although the identity of this channel is currently unclear.

Despite the assignment of IPC protection to a mitochondrial channel with K_ATP-like properties, the molecular identity of the channel remains unclear. Specifically, none of the KIR6.1/6.2 or SUR1/2A/2B proteins are known to contain mitochondrial targeting sequences [134,161–165]. Furthermore, although there are 14 KIR channel isoforms in mammals [105,166], the genetic model organism C. elegans contains only three such proteins (encoded by the irk-1,2,3 genes) [167]. In C. elegans with ablation of all three irk genes, no alteration in protection by hypoxic preconditioning or baseline sensitivity to hypoxic injury was seen [88]. Furthermore, mitochondria from these worms exhibited K⁺ channel activity with all of the pharmacologic properties of a K_ATP channel [88]. These data suggest that mitoK_ATP might not be a canonical KIR6/SUR channel.

Alternatively, another member of the Kir gene family, Kir1.1 (also known as renal outer medullary K⁺ channel, ROMK), has recently been proposed to encode a mitoK_ATP channel [95]. Specifically, ROMK variant 2 (ROMK2) contains an N-terminal mitochondrial localization sequence, and although its endogenous expression could only be detected by reverse-transcriptase polymerase chain reaction (RT-PCR), overexpression of recombinant ROMK2 fused to an epitope tag allowed immunodetection of its co-localization with mitochondrial markers. Recently, specific ROMK2 activators have been reported [168–170], but it is yet to be determined if these molecules can elicit protection against IR injury or activate a mitochondrial K⁺ flux. In addition, although a whole-body ROMK knockout exists [171,172], renal insufficiency and hypertension render this model unsuitable for cardiovascular studies such as IR injury, and as of the submission of the present study, a cardiac-specific ROMK knockout mouse has not been reported. Finally, the pharmacologic properties of the mitoK_ATP (e.g., sensitivity to DZX ATP, phosphadityl inositol bisphosphate (PIP₂), and fluoxetine) do not match those reported for ROMK (see ref. [159]). It is also intriguing that a recent abstract [173] claims to have identified the mitoK_ATP channel as a previously unknown protein (i.e., not ROMK). Hence, it seems prudent to keep an open mind as to whether ROMK2 is the bona fide mitoK_ATP channel.

Mitochondrial K_Ca2.x and K_Ca3.1 channels: composition, pharmacology, regulation, and role in IR protection

There are four genes in the Kcnn family. Kcnn1, Kcnn2, and Kcnn3, respectively, code for the SK potassium channels with IUPHAR names K_Ca2.1, K_Ca2.2, and K_Ca2.3 [174], while the Kcnn4 gene encodes an IK potassium (IK) channel termed K_Ca3.1 [43]. The K_Ca3.1 channel has not been implicated in IR protection; however, it does have a role in post-ischemic cardiac remodeling [175,176] and Kcnn4^−/− mice exhibit more damage in ischemic stroke [177]. The K_Ca2.x channels are expressed in atrial cells but not in ventricular tissue [178]. Their activation by DECBIO is protective against cardiac IR injury [179,180], and this protection is blocked by the K_Ca2.x antagonist NS8593. More recently, it has been claimed, on the basis of immunologic evidence, that the K_Ca2.x channels responsible for this cardioprotection are mitochondrial, and of the K_Ca2.2 and K_Ca2.3 variety [180], although similarity between these SK proteins precludes identification of the exact subtype.

K_Ca2.x channels are nominally activated by sub-micromolar Ca²⁺, co-ordinated by a calmodulin (CaM)-binding domain. Additional regulation of channel activity is also afforded by phosphorylation at the N- and C-termini [181–183]. These channels are also known to be activated by stimuli implicated in cardioprotection, such as 11,12-epoxyeicosatrienoic acid (EET) and NO^· [184,185]. However, both of these species can elicit protection via pleiotropic mechanisms [186,187], including other K_Ca channels [188–190]. The K_Ca2.x channels are selectively blocked by apamin, and are also nonselectively blocked by charybdotoxin (ChTx) [191] and fluoxetine, which also target BK and K_ATP channels, respectively [157,192]. This overlapping pharmacology with BK and K_ATP channels should be taken into consideration when interpreting pharmacologic evidence for a mitochondrial or cardioprotective role of K_Ca2.x channels. Ultimately, the use of Kcnn1–4^−/− mice [177,193,194] may be informative regarding the contribution of these channels to IR protection.

Mitochondrial K_Ca1.1 channels

K_Ca1.1 channels — composition

The K_Ca1.1 channel consists of a tetramer of Kcnma1 encoded pore-forming α subunits, each of which can be accompanied by a β subunit, with the entire complex also binding γ subunits. Each α subunit has seven trans-membrane helices, a voltage sensor, β/γ interaction domains, and a large cytosolic region containing two RCK (regulation of conductance of K⁺) domains which house the Ca²⁺ sensing ‘bowl’. The exon 19–23 region of Kcnma1 (between the RCK domains) can be alternatively spliced (Figure 4) [48,195–197], yielding isoforms termed ‘zero’ (no exons 19–23), ‘e20’ (IYF insert between 19 and 20) [198], ‘e21/STREX’ (59 AA insert in exon 21) [198–201], ‘e22’ (inclusion of exon 22) [198], ‘Δe23’ (loss of exon 23) [202,203], and ‘DEC’ (C-terminal splice variant) [204]. Notably, the DEC variant has been suggested to impart mitochondrial localization [204]; however, this moiety has also been observed to increase surface K_Ca1.1 expression in combination with a β4 subunit [205,206].

Figure 4. Venn diagram representing the various K⁺ channel channels, isoforms, and their subgroupings.

The 6-transmembrane (6TM) pore-forming core is represented as a rounded rectangle with the individual N- and C-termini and their corresponding unique exons/domains illustrated as darker shaded shapes (see key). As only gross structural motifs are depicted in this figure, specific differences in amino acid sequence (i.e., differences in charged residues in S4 between K_Ca1.1 and K_Na1.x or unique sequences in K_Na1.1 and K_Na1.2 RCK domains) are not represented.

There are four genes encoding K_Ca1.1 β subunits (Kcnmb1–4) [207,208] and their expression is tissue-specific [208,209]. These proteins interact with K_Ca1.1-α, altering activity and drug sensitivity [209–214]. β1 and β2 both increase Ca²⁺ sensitivity [213–216], and β1 also slows activation/inactivation kinetics [213,217] and affects cellular localization [215]. β2 and some splice variants of β3 also contain globular N-terminal domains, which confer rapid ‘N-type inactivation’ to K_Ca1.1 currents [211,214,218–222], whereas β4 down-regulates channel activity [223]. Mitochondrial localization of the β subunits, particularly β1 [189], β3 [82], and β4 [189], has been reported; however, these reports are largely informed by co-immunoprecipitation of the β subunits with the α subunit from mitochondrial-enriched tissue preparations. Currently, there are no data demonstrating that the β subunits complex with the α subunits in mitochondria and affect the channels’ pharmacologic or electro-physiologic properties. K_Ca1.1 γ subunits are single transmembrane leucine-rich repeat proteins encoded by the Lrrc26 (γ1), Lrr52 (γ2), Lrr55 (γ3), and Lrr38 (γ4) genes. The γ subunit interaction with the extracellular face of K_Ca1.1-α lowers channel voltage sensitivity [224,225]. Although γ subunits are diversely expressed, to date none have been detected in the heart [226].

Knockout mice exist for the genes encoding several components of the K_Ca1.1 channel, including the Kcmna1 [90,227], Kcnmb1 [220], and Kcnmb4 [228]. These mice exhibit a variety of phenotypes including spontaneous death, motor dysfunction, circadian rhythm disruption, and vasoconstriction, although this has not precluded their use in studying the role of K_Ca1.1 in cardioprotection (see below). Functional insight has also been afforded by K_Ca1.1 channel crystal structures [229,230].

K_Ca1.1 channels — pharmacology and regulation

There are no drugs that distinguish K_Ca1.1-α splice variants, although β/γ subunits have been shown to affect K_Ca1.1 pharmacology. The peptide toxins ChTx, iberotoxin (IbTx), and slotoxin (SloTx) all occlude the pore on the outer face of the channel [231–233] and are useful for measuring surface K_Ca1.1 function, but their membrane impermeability renders them unsuitable for probing intracellular K_Ca1.1 activity. Sensitivity to ChTx is increased 20-fold by β1 [231], whereas sensitivity to SloTx is decreased by β1 or β4 [234], and inhibition by ChTx or IbTx is lost in the K_Ca1.1-α/β4 composition [235,236]. Although the molecule rottlerin (also known as mallotoxin, historically thought to be a PKC inhibitor [237]) is known to activate K_Ca1.1 [238], γ1 subunit-containing channels are resistant to such activation [239,240]. The small-molecule paxilline is reportedly a membrane-permeable K_Ca1.1 blocker, but its specificity has been questioned by the reported efficacy in Kcnma1^−/− mice [97]. The neurosearch (NS) class of compounds (NS004, NS1619, and NS11021) was developed as K_Ca1.1-specific activators [241,242] and is cardioprotective (Figure 2 and Table 1) [96,243–245], but also exhibits multiple K_Ca1.1-independent effects [246–253]. This includes inhibition of SERCA [254], L-type Ca²⁺ channels [246], Ca²⁺-activated Cl⁻ and Na⁺ channels [247], and Ca_V channels [248,255,256]. In addition, these drugs activate K_V7.4 and nAchR α7, and are known to have multiple effects on mitochondrial function [98,252,257]. Finally, emodepside has been reported to activate K_Ca1.1 at nM concentrations [258,259]. Overall consensus is that peptide-based K_Ca1.1 modulators are specific but of limited in situ utility, whereas current small-molecule K_Ca1.1 modulators are less specific but more useful in a variety of cell, organ, or in vivo settings.

In addition to direct pharmacology, many signaling pathways are known to endogenously regulate K_Ca1.1 channels in a cellular context. These include arachidonic acid metabolism (i.e., EETs) [189,260], NO^· metabolism [261,262], pH [263], Zn²⁺ homeostasis [264], phosphorylation by PKA, PKG, PKC, and CaMKII [265–268], and palmitoylation and myristoylation [269–271]. Unfortunately, given the enormous scope of this subject, full discussion of these signaling pathways and their context in IR injury and protection is not possible in the current review. Finally, given that K_Ca1.1 is a K_Ca channel, it should be noted that modulation of intracellular Ca²⁺ by other Ca²⁺ channels (e.g., ryanodine and IP₃ receptors) can also affect its activity [272–275].

Mitochondrial K_Ca1.1 channels — role in IR protection

The first report of a mitochondrial large-conductance (295pS) K⁺ channel employing patch clamp of glial cell mitochondrial inner membranes [81] showed that the channel was activated by Ca²⁺ (0.1–1 μM) and voltage, and blocked by ChTx. A large-conductance K⁺ (BK) channel opened by NS1619 was also found in isolated liver mitochondria, and this compound was protective in a rabbit heart model of IR injury, in a manner blocked by paxilline [42]. Further work confirmed the cardioprotective nature of the NS compounds and also reported on a potential role for a mitochondrial BK channel in cardioprotection by volatile APC [74,97,242]. Specifically, the K_Ca1.1 antagonist IbTx was reported to block volatile anesthetic or NS protection in a model of ischemic postconditioning [276]. These studies assigned a variety of names to the mitochondrial BK channel including BK, K_Ca, and BK_Ca (see Section ‘Mitochondrial K⁺ Homeostasis and Discovery of Mitochondrial K⁺ Channels’), although a specific gene or protein name was conspicuously absent. Notably, cardioprotection induced by the K_Ca1.1-activating NS compounds was not blocked by the mitoK_ATP antagonist 5-HD [125]. In addition, the mitochondrial BK channel itself was found to be insensitive to DZX and 5-HD [277], suggesting that this channel is distinct from mitoK_ATP. Later, immunologic studies documented K_Ca1.1 expression in the heart [196,198,207], and it was found that K_Ca1.1 channels are not expressed at the myocyte plasma membrane or sarcoplasmic reticulum membrane [42,277]. Furthermore, K_Ca1.1 was found specifically in mitochondrial membranes [42,81,204,278,279], including the presence of the β1 subunit [41,42,278,280,281]. Together, these studies suggested the existence of a bona fide K_Ca1.1 channel in the mitochondrial inner membrane, with a proposed role in mediating the cardioprotective effects of APC.

Despite these early studies, the case for a role of K_Ca1.1 in cardioprotection by APC has previously been challenged by our finding that APC protection was intact in Kcnma1^−/− hearts [97]. Thus, K_Ca1.1 is dispensable for APC. Furthermore, as detailed above, the specificity of the NS compounds for K_Ca1.1 has been repeatedly questioned [119,246–253,257]. For example, the PKA inhibitor H-89 blocks APC protection [282], but does not block protection afforded by NS1619 [41], thus suggesting that NS compounds and APC protect via different mechanisms. In general, the NS compounds are unsuitable for drawing conclusions about the molecular identity of channels that underlie APC.

This concept is further illustrated by the finding that while IR protection by NS compounds was absent from Kcnma1^−/− hearts, it was still present in Kcnma1^−/− cardiomyocytes. This suggests that a K_Ca1.1-independent side effect of the NS compounds was responsible for their protective effects in the isolated cell system [98], but for poorly understood reasons this K_Ca1.1-independent side effect was not able to be recruited for protective benefit in the intact heart. Further experimentation revealed a role for K_Ca1.1 channels within intrinsic cardiac neurons (where K_Ca1.1 is known to be expressed [283,284]) in mediating the protective effects of NS compounds in the intact heart [98]. Overall, it is suggested that APC cardioprotection does not require K_Ca1.1, whereas NS cardioprotection requires K_Ca1.1 in a noncardiomyocyte cell (probably cardiac neurons). It should be noted that our data do not preclude the possibility that a mitochondrially localized K_Ca1.1 channel exists and may indeed be a viable drug target to induce cardioprotection. However, such a channel has no role in APC and cannot be inferred from the use of NS compounds.

Finally, although the use of nonspecific K_Ca channel inhibitors has led to suggestions of a potential role for mitochondrial K_Ca1.1 channel in IPC [124,125], such a case is refuted by our finding that protection induced by IPC was completely intact in Kcnma1^−/− hearts [98]. Thus, without a defined role in IPC or APC, any potential cardioprotective effects of a mitochondrial K_Ca1.1 channel are limited to specific agonists that are yet to be discovered.

Mitochondrial K_Na1.x channels

K_Na1.x channels — composition

The K_Na1.x channels consist of tetramers of pore-forming α subunits encoded by either Kcnt1 or Kcnt2. These genes are classified as being in the Slo gene family, despite having only 7% homology to Kcnma1. K_Na1.x channels lack the S0 domain of K_Ca1.1 and are therefore unable to interact with β subunits [50]. They also lack the voltage-sensing positive residues in transmembrane S4. K_Na1.x and K_Ca1.1 also differ in their cytosolic domains, with K_Na1.x RCK domains activated by Na⁺ and Cl⁻ and inhibited by Ca²⁺ (Figure 4) [285]. The K_Na1.1 paralog has five splice variants (termed Slack-A, Slack Ax2, Slack B, Slack Bx2, and Slack M [286]) in the cytosolic N-terminal region, of which two have been characterized: Slack-B is the canonical K_Na1.1 channel, whereas Slack-A produces a channel with properties similar to K_Na1.2 [49,55,286]. While no splice variant of K_Na1.2 has been discovered, K_Na1.2 can heteromultimerize with Slack-B [55] or with K_Ca1.1, to produce channels with novel characteristics [50]. The K_Na1.x channels have been mainly studied in the brain where they are highly expressed, and currently, only K_Na1.2 has been detected in the heart [51,89,286,287]. This is consistent with early observations of surface K_Na channels in the heart [288].

The K_Na1.2 channel is unique among channels encoded by the Slo gene family, in that it has a nucleotide-binding domain on the C-terminus (Figure 4). ATP binding to this domain inhibits the channel in rodent cells [49], but activates in human and frog cells [287,289–297]. Meanwhile, NAD⁺ and NADP⁺ both activate K_Na1.2 presumably via this nucleotide-binding domain [298], but it should be noted that pyridine nucleotides can also act on K_ATP [299] and K_Ca channels [300,301], so this property cannot be used to distinguish these channels. Interestingly, in neurons of both humans and rats, K_Na1.2 transcription is regulated by nuclear factor kappa B (NF-κB) activation [302], which itself is activated by IPC and APC [303–306]. Knockout mice for Kcnt1, Kcnt2, and a double knockout are available and have been used to determine the role of these channels in cardioprotection (see below) [89,307].

K_Na1.x channels — pharmacology and regulation

K_Na1.x channel pharmacology is largely informed by K_Na1.1, although several of these drugs are also known to affect the lone C. elegans SLO2 channel [97], and thus, not surprisingly, most K_Na1.x drugs exhibit similar effects on both mammalian paralogs. There are numerous drugs that nonspecifically inhibit both K_Ca1.1 and K_Na1.x channels, including paxilline, verapamil, and bepridil [308,309]. Similarly, several drugs can nonspecifically activate both K_Ca1.1 and K_Na1.x channels: bithionol [310–312], 17β-estradiol [313,314], and nonsteroidal anti-inflammatory fenmates such as niflumic acid [294]. In addition, several drugs that target other K⁺ channels do not affect K_Na1.x channels, including (target) dendrotoxin (K_V), apamin (K_Ca2.x), glibenclamide (K_ATP), or DZX (K_ATP). This can help in ruling out a role for K_Na channels in any particular phenomenon [49,50].

In terms of distinguishing K_Ca1.1 from K_Na1.x, ChTx and IbTx have already been described above as K_Ca1.1-specific inhibitors, and NS1619 and EtOH as K_Ca1.1 activators; none of these agents affect K_Na1.x [315]. There are also inhibitors of K_Na1.x including clofilium and quinidine [290,308,309,316] and activators of K_Na1.x including loxapine and niclosamide (see Table 1) [317]. None of these agents have an impact on K_Ca1.1. Unfortunately, many of these drugs also affect other channels, including K_Ca5.1 (also known as SLO3, encoded by Kcnu) [318], K_V11 (human ether-a-go-go related gene) [319], K_V10 (human ether-a-go-go potassium channel) [320], K_V1.5 [321], K_V7 (Kcnq) [322], and K_V4 (Kcne) [323] channels. As such, the use of pharmacologic agents to assign distinct functions to K_Na1.x channels is susceptible to a multitude of off-target effects.

In addition to pharmacologic regulation, several signaling pathways are known to regulate K_Na1.x channels either directly or at the transcriptional level. NAD⁺ regulates both K_Na1.2 [298] and K_Na1.1, resulting in activation and a lower EC₅₀ for Na⁺. In the brain, K_Na1.2 channels are inhibited by PKC, G_α, G_q, M1, and mGluR1 receptor signaling pathways, whereas K_Na1.1 channels are activated by these same pathways [55].

K_Na1.x channels — role in mitochondria and cardioprotection

In isolated cardiac mitochondria, functional assays (Tl⁺ flux) have demonstrated a channel with BK-like pharmacologic sensitivity (i.e., activation by bithionol or isoflurane and inhibition by bepridil) that is absent from mitochondria from Kcnt2^−/− mice [307]. This channel was still present in Kcnt1^−/− mice and absent from Kcnt1/Kcnt2 double knockouts, supporting the existence of a K_Na1.2 channel in heart mitochondria [307]. To date, immunologic and other approaches have not yielded solid evidence for the mitochondrial K_Na1.2 channel, primarily due to issues of antibody specificity (unpublished observations), although this is a problem not unique to K_Na1.2 (see discussion above on mitoK_ATP and K_Ca1.1 channels). Electrophysiology (patch clamp) studies on mitoplasts from wild-type and Kcnt2^−/− mice are currently underway in our laboratory.

Both Kcnt2^−/− and Kcnt1^−/− mice retain cardioprotection by IPC and DZX, consistent with the action of a mitoK_ATP channel and not a K_Na1.x channel in such protection. In addition, the Kcnt1^−/− heart was protected by APC (via isoflurane). However, in Kcnt2^−/− and in Kcnt1/Kcnt2 double knockout hearts, no protection by APC was observed [307]. These data suggest an absolute requirement of K_Na1.2 for the protective effects of APC. Furthermore, the K_Na1.x activator bithionol was found to be protective when delivered exogenously, supporting that opening of K_Na1.x alone is sufficient to confer IR protection.

The single C. elegans SLO2 channel is also required for hypoxic protection. Specifically, protection against anoxia–reoxygenation injury by the volatile anesthetic isoflurane was lost in Slo2^−/− worms. In addition, mitochondria from Slo2^−/− worms lacked a BK-like channel activity seen in wild-type mitochondria (i.e., activated by bithionol or isoflurane, blocked by bepridil, and insensitive to IbTx) [97]. Taken together, these knockout organism experiments demonstrate that a channel with BK-like activity in the mitochondrion (SLO2 in worms and K_Na1.2 in mammals) is a conserved mechanism for protection against IR injury triggered by APC.

In the context of ion homeostasis in IR injury and cardioprotection, there are clear reasons for hypothesizing that both Na⁺-activated and Ca²⁺-activated mitochondrial channels (i.e., K_Na1.2 and K_Ca1.1, respectively) may be opened by the high prevailing concentrations of Na⁺ or Ca²⁺ during ischemia (see Section ‘Ischemia–Reperfusion Injury and Protection’) [51,324–326]. However, evidence for the opening of these channels in baseline IR injury alone is lacking, and blockers of these channels do not exacerbate IR injury [92,307]. In addition, how the levels of Na⁺ and Ca²⁺ in the heart differ in ischemia following either IPC or APC is poorly understood. There is currently no rationale for the hypothesis that these channels would open in response to their natural ligands (Na⁺ or Ca²⁺) under cardioprotective stimulus conditions. In the case of K_Na1.2, elevated Ca²⁺ would be predicted to inhibit the channel despite the raise in Na⁺ [50,285]. The likely activators of these channels in IPC are upstream protein kinase signaling pathways (see above), while in the case of APC it is likely that volatile anesthetics are direct channel ligands.

While the use of volatile anesthetics is positively linked with reduced mortality in cardiac surgery [327], there is also evidence that repeated exposures can result in acute hepatitis [328]. Therefore, the identification of the channel that underlies the clinically important phenomenon of APC potentially paves the way for the development of novel K_Na1.2-targeted cardioprotective therapeutics [329]. However, an important caveat to these results is that we are the only laboratory that has to date investigated or provided any evidence for mitochondrial K_Na1.2. As such, validation of these findings by other laboratories will be necessary before moving toward any potential clinical applications.

Downstream mechanisms of protection due to mitochondrial K⁺ channel opening

The mitochondrial PT pore is a fundamental arbiter of cell survival in IR injury (Figure 1) [71,330–334]. As such, numerous events at the mitochondrial level that are known to regulate the PT pore (e.g., mitochondrial Ca²⁺ overload, ROS generation, energetics, and pH) have been shown to interface with upstream signaling pathways implicated in cardioprotection (GSK-3B, NO^·, signaling ROS, PKA, PKC, and others) [335–339]. However, despite a proposed central role for mitochondrial K⁺ channels in IR protection (see Section ‘Mitochondrial K⁺ Homeostasis and Discovery of Mitochondrial K⁺ Channels’), how such channels elicit downstream protective mitochondrial events is poorly understood.

There are numerous attractive hypotheses linking mitochondrial K⁺ channels to the PT pore, and these can be roughly broken down into those dependent on membrane potential, and those that are not. Owing to the mitochondrial K⁺ cycle (Figure 2), it is apparent that opening of a mitochondrial K⁺ channel may serve (coupled with a KHE) to uncouple mitochondrial oxidative phosphorylation. Mild uncoupling of Ox-Phos alone is known to be cardioprotective [22,340–342] and may have many salutary effects on IR injury, such as those described in the following paragraphs.

ROS generation

Tissue reperfusion following ischemia is known to trigger a burst of ROS, and it is also known that mitochondrial ROS generation is exquisitely sensitive to membrane potential [77,78]. As such, it has been proposed that mild uncoupling may serve to depress ROS generation in early reperfusion [343–345]. This may be achievable via opening of a mitochondrial K⁺ channel [244,245,346–348]. In addition, APC protection may also decrease ROS at reperfusion via mild uncoupling [349–351]. These findings would appear to position mitochondrial K⁺ channel opening upstream of a decrease in pathologic ROS.

However, the interplay between ROS and mitochondrial K⁺ channels in the setting of IR injury is much more complicated. Specifically, it is well known that low levels of ROS (termed ‘signaling ROS’) are in fact required for the cardioprotective effects of IPC [352–356] and APC [357]. This is consistent with the notion that ROS is hormetic, and indeed, low levels of ROS alone are known to confer IR protection [353,358,359]. In addition, the mitoK_ATP channel is redox-sensitive and opens in response to a variety of ROS [360]. These findings position mitochondrial K⁺ channel opening downstream from signaling ROS.

Still further complication arises from the claim that opening of a mitochondrial K⁺ channel itself can trigger ROS generation by complex I [361], which would position channel opening upstream of signaling ROS. Overall, it appears that the relationship between mitochondrial K⁺ channels and ROS generation may be bi-directional, with signaling ROS and mitochondrial K⁺ channel opening perhaps exhibiting an amplification loop behavior during the early trigger phase of cardioprotection, leading to an overall decrease in pathologic ROS at reperfusion. Unfortunately, beyond the brute-force application of antioxidants, the evidence for a role of ROS in transmitting a protective signal as part of an IPC or mitochondrial K⁺ channel signaling cascade is somewhat limited. Clearly, much remains to be done, in elucidating the order of events relating mitochondrial K⁺ channels and ROS in IR protection. Recent progress in identifying the molecular constituents of mitochondrial K⁺ channels (see Sections ‘Mitochondrial K_ATP Channel: Composition, Pharmacology, Regulation, Role in IR Protection; Mitochondrial K_Ca2.x and K_Ca3.1 Channels: Composition, Pharmacology, Regulation, Role in IR Protection; Mitochondrial K_Ca1.1 Channels and Mitochondrial K_Na1.x Channels; and Mitochondrial K_Na1.x Channels’) may also permit the identification of redox-sensitive residues (e.g., cysteine and methionine) within these proteins that are responsible for the interplay of these channels with ROS.

Mitochondrial Ca²⁺ and autophagy

Another potential benefit of mild uncoupling via opening of a mitochondrial K⁺ channel would be the prevention of mitochondrial Ca²⁺ uptake, which is driven by the membrane potential. As such, opening of a mitochondrial K⁺ channel may prevent mitochondrial Ca²⁺ overload [362,363], thus serving to prevent PT pore opening. A third potential benefit of mild uncoupling downstream from mitochondrial K⁺ channel opening could be the triggering of mitophagy [364], which is itself known to be cardioprotective [342,365]. The ability of mitochondrial K⁺ channel opening to regulate mitophagy has not been rigorously investigated, although it was shown that the mitoK_ATP opener DZX induces mitophagy in murine hearts [366], and it is also known that isoflurane induces cardiac mitophagy [367].

In considering the above phenomena, a caveat should be rendered regarding any link between mitochondrial K⁺ channel opening and mitochondrial uncoupling, in terms of the size of the K⁺ conductance involved. Specifically, it is known that opening of the mitoK_ATP channel only drops membrane potential by 1–2 mV in isolated cardiac mitochondria (from its baseline value of ~180 mV) [368]. Thus, any such uncoupling mediated by a mitoK_ATP channel is likely to be insufficient to affect mitochondrial function (i.e., ATP production), although it could be sufficient to affect the driving force for mitochondrial ion fluxes. In contrast, the larger conductance of mitochondrial BK channels renders them more attractive candidates for inducing uncoupling. In this regard, we have found that the K_Na1.x opener biothionol (which is cardioprotective, [307]) is also capable of inducing mitochondrial uncoupling in cardiomyocytes (unpublished data). In addition, the K_Na1.x opener niclosamide was recently reported to uncouple mitochondria [369]. It is not yet known if the uncoupling effect of niclosamide is mediated by a mitochondrial K⁺ channel or is capable of conferring cardioprotection.

Membrane potential-independent effects and volume

Beyond effects that depend on membrane potential, mitochondrial K⁺ channels are thought to play a role in regulating mitochondrial volume [74] (see Section ‘Mitochondrial K⁺ Homeostasis and Discovery of Mitochondrial K⁺ Channels’ and Figure 2). Thus, it is possible that mild swelling associated with mitochondrial K⁺ channel opening [370] may be part of a protective signaling cascade. Mitochondrial volume has been historically linked to respiratory function, with the transition between classical respiratory state 4 (quiescent) and state 3 (phosphorylating) being associated with a contraction of the mitochondrial matrix. [75,76,371]. As such, mild swelling would be expected to coincide with a lower overall mitochondrial respiratory function. How this would lead to protection against IR injury is not clear.

Alternatively, mitochondrial swelling could confer protection by many other mechanisms as follows: (i) by improving efficiency of the creatine kinase energy shuttle, for example, by changing the distance between inner and outer mitochondrial membranes. (ii) By regulating the supra-molecular assembly of respiratory chain complexes and super-complexes [372]. For example, it has been shown that mitochondria from hearts protected by APC had improved ATP synthase function [335]. Furthermore, it has recently been proposed that the cardio-protective drug SS-31 (Bendavia) [373] may confer protection via the stabilization of super-complexes involving cardiolipin [374–376]. (iii) Mild swelling could interfere with PT pore assembly [332,377]. (iv) Mild swelling would also be expected to dilute the contents of the mitochondrial matrix, which may directly affect the activity of enzymes in the tricarboxylic acid cycle by lowering substrate concentrations, or may affect concentrations of important mitochondrial enzyme allosteric regulators such as Ca²⁺, NADH, acetyl-CoA, and phosphate. (v) By physiologic coupling to other mitochondrial channels or transporters that can sense volume or osmolarity.

From a perspective of long-term protective benefits of mitochondrial K_ATP opening, it has been shown that the treatment of cells with DZX causes mild in situ mitochondrial swelling, which can trigger a signaling cascade involving cyclic AMP responsive binding element (CREB) and NF-κB, leading to resistance to apoptosis [378]. Thus, there are clearly cell signaling mechanisms triggered by mitochondrial volume changes, which may play an important role in IR protection and remain to be determined.

In summary, the events linking mitochondrial K⁺ channel opening to protection from IR injury are currently poorly understood, both at the mechanistic and molecular levels. It is hoped that the future availability of specific mitochondrial K⁺ channel ligands (facilitated by the molecular identification of these channels) will permit the independent interrogation of mitochondrial K⁺ channel opening and swelling as a signaling trigger mechanism, to elucidate these downstream pathways.

Physiologic role of mitochondrial K⁺ channels beyond cardioprotection

Given the importance of K⁺ as a cytosolic solute and the conserved nature of mitochondrial K⁺ channels, it is important to consider the endogenous physiologic role(s) of these channels in the cell, beyond protection from IR injury. Such considerations could also provide insights into novel mechanisms of regulating mitochondrial function.

At the organism level, as already discussed in Section ‘Mitochondrial K_Ca2.x and K_Ca3.1 Channels: Composition, Pharmacology, Regulation, Role in IR Protection’, mammalian K_ATP channels play important roles in glucose-stimulated insulin secretion in pancreatic β-cells [379], and these channels are the pharmacologic target of the widely used antidiabetic sulfonylurea class of drugs. In addition, mammalian BK channels are broadly recognized to play roles in regulating vascular smooth muscle tone [380,381], in muscle relaxation [382], in regulating circadian rhythms [383,384], and in the function of neurons in the dorsal root ganglion [89]. Both K_Na1.x paralogs are highly expressed in the brain, and the majority of research on endogenous K_Na1. x channels has been conducted in neurons, focused mainly on K_Na1.1 [287]. Whether any of these functions attributed to K_Na1.x channels are in fact due to such channels located in mitochondria is not known.

Evidence for a direct physiologic function of mitochondrially localized K⁺ channels beyond their role on protection against IR injury is very sparse. Historically, uncoupling has been viewed as an important contributor to basal metabolic rate, perhaps best envisioned in brown adipose tissue (BAT), which burns fat to generate heat via mitochondrial uncoupling. The discovery of homologs of the BAT uncoupling protein (now called UCP1, [385]) in other tissues has led to a consensus that these proteins (UCP2–5, [386–389]) may serve a role in regulating whole-organism energy expenditure [390–392]. An alternative uncoupling mechanism has also recently been proposed, involving a nonspecific pore formed by the c-subunit of the ATP synthase in mitochondria [393]. Whether uncoupling by opening of a mitochondrial K⁺ channel (presumably of the K_Na1.x variety — see Section ‘Downstream Mechanisms of Protection by Mitochondrial K⁺ Channel Opening’) is capable of having a similar effect on whole-organism energy balance remains to be seen, although it is exciting that the K_Na1.x activator niclosamide is reported to have an antiobesity effect similar to the uncoupler 2,4-dinitrophenol [369]. This result potentially positions mitochondrial K_Na1.2 as a candidate antiobesity drug target.

Outlook

In the roughly three decades, since the discovery of IPC [394] and APC [395], there has been a plethora of research devoted to understanding the molecular underpinnings of these phenomena. Although mitochondrial K⁺ channels were identified as candidate players early in this research arc, only in the past 4 years have viable molecular identities been assigned to these channels: K_Na1.2 [307], K_Ca2.2/K_Ca2.3 [180], and KIR1.1 (with caveats as outlined in Section ‘Mitochondrial K_ATP Channel: Composition, Pharmacology, Regulation, Role in IR Protection’) [95]. These identities can now be used to develop novel molecules to afford protection of organs such as the heart and brain from ischemic injury in a clinical setting.

Finally, it is noteworthy that the field of mitochondrial K⁺ channel research has used model organisms at multiple stages, including genetically engineered mice, C. elegans, plants [138], amoeba [139], and trypanosomes. In addition, the field exists as a clear demonstration of the importance of basic biomedical research toward understanding a clinically relevant phenomenon in humans. The discoveries made regarding mitochondrial K⁺ channels in the past 4 years provide a rich resource for future development of clinical therapies.

Abbreviations

5-HD

5-hydroxydecanoate

APC

anesthetic preconditioning

BAT

brown adipose tissue

BK

big conductance of potassium

CaMKII

Ca²⁺/calmodulin-dependent protein kinase II

ChTx

charybdotoxin

DZX

diazoxide

EET

eicosaepoxytrienoic acid

IbTx

iberotoxin

IK

intermediate conductance

IPC

ischemic preconditioning

IR

ischemia–reperfusion

IUPHAR

International Union of Basic and Clinical Pharmacology

K_Ca1.1

channel encoded by Kcnma1, also known as SLO1

K_Ca2.1

channel encoded by Kcnn1, also known as SK1

K_Ca2.2

channel encoded by Kcnn2, also known as SK2

K_Ca2.3

channel encoded by Kcnn3, also known as SK3

K_Ca3.1

channel encoded by Kcnn4, also known as IK, SK4

K_Ca5.1

channel encoded by Kcnu, also known as SLO3

KHE

K⁺/H⁺ exchanger

KIR

inwardly rectifying potassium channel

K_Na1.1

channel encoded by Kcnt1 (formerly Slo2.2), also known as Slack, K_Ca4.1, SLO2.2

K_Na1.2

channel encoded by Kcnt2 (formerly Slo2.1), also known as Slick, K_Ca4.2, SLO2.1

mitoK_ATP

ATP-sensitive mitochondrial potassium channel

NCX

sodium/calcium exchanger

NHE

sodium/proton exchanger

NO^·

nitric oxide

PKA

cAMP-dependent protein kinase

PKC

Ca²⁺/diacylglycerol-dependent protein kinase

PKG

cGMP-dependent protein kinase

PT

permeability transition

RCK

regulation of conductance of K⁺

ROMK

renal outer medullary potassium channel

ROS

reactive oxygen species

SERCA

sarco/endoplasmic reticulum calcium-ATPase

SK

small conductance

SLO

slowpoke

Slo2

C. elegans gene encoding the single isotype SLO2 K_Ca channel

SloTx

slotoxin

SUR

sulfonylurea receptor