Two-Pore-Domain Potassium (K2P-) Channels: Cardiac Expression Patterns and Disease-Specific Remodelling Processes

Two-pore-domain potassium (K2P-) channels conduct outward K+ currents that maintain the resting membrane potential and modulate action potential repolarization. Members of the K2P channel family are widely expressed among different human cell types and organs where they were shown to regulate important physiological processes. Their functional activity is controlled by a broad variety of different stimuli, like pH level, temperature, and mechanical stress but also by the presence of lipids or pharmacological agents. In patients suffering from cardiovascular diseases, alterations in K2P-channel expression and function have been observed, suggesting functional significance and a potential therapeutic role of these ion channels. For example, upregulation of atrial specific K2P3.1 (TASK-1) currents in atrial fibrillation (AF) patients was shown to contribute to atrial action potential duration shortening, a key feature of AF-associated atrial electrical remodelling. Therefore, targeting K2P3.1 (TASK-1) channels might constitute an intriguing strategy for AF treatment. Further, mechanoactive K2P2.1 (TREK-1) currents have been implicated in the development of cardiac hypertrophy, cardiac fibrosis and heart failure. Cardiovascular expression of other K2P channels has been described, functional evidence in cardiac tissue however remains sparse. In the present review, expression, function, and regulation of cardiovascular K2P channels are summarized and compared among different species. Remodelling patterns, observed in disease models are discussed and compared to findings from clinical patients to assess the therapeutic potential of K2P channels.


Introduction
Two-pore-domain potassium (K 2P ) channels are expressed throughout the human body and contribute to background potassium conductance in many different cell types [1,2]. In the human genome 15 K 2P channels have been described which differ from classical potassium channels by the fact that each subunit carries two pore-forming domains, and the channels thus assemble as dimers instead of tetramers ( Figure 1). K 2P channels give rise to background or "leak" potassium currents which control a multitude of physiological processes [1]. Initially, K 2P currents were described as outward rectifying "leakage currents" but recent work has shown that several members of the K 2P family can also be voltage activated [3].
K 2P currents show a high degree of similarity to the potassium plateau currents I KP , described in guinea-pig cardiomyocytes and the steady-state potassium current I K,SS, characterized in murine cardiomyocytes and the arachidonic acid-sensitive potassium current I KAA from rat ventricular cardiomyocytes [4][5][6][7]. Cardiac mRNA abundance was described for several members of the K 2P family (Figure 2) In the present review, expression, function, and regulation of cardiovascular K 2P channels are summarized and compared among different species. Remodelling patterns, observed in disease models are discussed and compared to findings from clinical patients to assess the therapeutic potential of K 2P channels ( Figure 3).  Cardiac mRNA levels of K 2P channels in the human heart (whole tissue). Expression of two-pore-domain potassium (K 2P -) channel mRNA level in human right atrial (n = 10) and left ventricular (n = 5) tissue samples. Data are given as mean ± SEM relative to the housekeeping gene importin 8 (IPO8). * indicate p < 0.05 from Student's t-tests. Data from Schmidt et al. 2015, Circulation [8].

Structural Assembly and Nomenclature of K 2P Channels
The 15 channel subunits of the K 2P family each consists of around 300 and 550 amino acids. The sequence differences between the individual subunits of the K 2P channel can sometimes be as large as to other potassium channel families. K 2P 18.1 (TRESK) channel subunits, for example share only 19% amino acid sequence identity with the other K 2P family members. But the common feature that links them is the eponymous structural motif of two pore-forming domains per subunit, which distinguishes them from all other potassium channel groups. As shown in Figure 1, the four alphahelical transmembrane domains (M1-M4) flank two pore-forming loops (P1 and P2), each containing the potassium selective filter motif (GLG, GFG, or GYG). M1 and P1 are connected by a long extracellular loop, forming an overhead cap structure. The short amino terminus and a much longer carboxy terminus, which contains a variety of regulatory phosphorylation and protein interaction motifs, are localized intracellularly. Whereas most potassium channels form tetramers with one pore-forming loop per subunit, a functional two-pore domain potassium channel is composed of two alpha subunits ( Figure 1). In addition to homodimerization, certain K 2P channel subunits can also assemble as heterodimers. This is mainly described within the same subfamilies (i.e., TASK-1/TASK-3, TREK-1/TRAAK, THIK-1/THIK-2), but can also occur between TWIK-1 and TREK or TASK-1, and between TASK-1/TALK-2 subunits. Physiological relevance in the perception of hypoxia has been described for TASK-1/TASK-3 heterodimers and TWIK-1/TREK-1 heterodimers have been detected in astrocytes. Apart from the TASK-1 and TALK-1 subfamilies, all K 2P channel subunits possess a conserved Cys-amino acid residue of the overhead domain that is thought to play a major, although not yet conclusively elucidated, role in dimerization. The predicted membrane topology and tertiary structure have already been confirmed by X-ray structural analysis for several K 2P -channels (Table 1). The name kcnk8 was initially given to a murine K 2P gene which was later identified as an ortholog of the human KCNK7 gene and therefore renamed to kcnk7 KCNK9 K 2P 9.1 TASK-3 (TWIK-related acid-sensitive K + channel 3) KT3.2, BIBARS, TASK32 -  Upon their discovery, the individual K 2P -channels received trivial names reflecting their respective structural and regulatory properties: TWIK: "Tandem of P domains in a weak inward rectifying K + channel", TREK: "TWIK-related K + channel", TASK: "TWIKrelated acid-sensitive K + channel", TRAAK: "TWIK-related arachidonic acid activated K + channel", TALK "TWIK-related alkaline pH-activated K + channel", THIK "tandem pore domain halothane-inhibited K + channel", and TRESK "TWIK-related spinal cord K + channel". In parallel, the channels are labeled consecutively with the designations K 2P 1.1 to K 2P 18.1 according to the Human Genome Organization name of the encoding gene (KCNK1 to KCNK18) (see Figure 2 and Table 1). Each of the 15 subfamilies members (K 2P 1.1 to K 2P 18.1) contains only one member. Unfortunately, this led to a confusing nomenclature in which channels with different functional properties such as TASK-1 and TASK-2 have similar names, while other channels are titled with acronyms of factually incorrect names (for example, TWIK-1 is not a weak inward rectifier but an open rectifier). Further, some channels carry a variety of redundant names such as in case of K 2P 3.1: TBAK1, TASK1 and OAT1. Several KCNKx designators were initially assigned to K 2P channel transcripts that later turned out to be orthologs of other human K 2P channels. Thus, KCNK8 (the murine transcript designated kcnk8 later proved to be an ortholog of human KCNK7 and was therefore renamed kcnk7), KCNK11, and KCNK14 (both orthologs of KCNK15) do not exist [9]. For better understanding, we will provide the trivial names of the channel subunits in brackets in addition to the International Union of Basic and Clinical Pharmacology IUPHAR (K 2P X.1) names. Since they do not show any functional activity in heterologous expression systems, the channels KCNK7, KCNK12 and KCNK15 are referred to as silent K 2P channels. It remains unclear whether these K 2P channel subunits are truly nonfunctional in vivo or whether they just lack essential cofactors to achieve functionality upon heterologous expression. In fact, the functionality of the K 2P 16.1 channels could be restored by deletion of an n-terminal ER-retention motif [8].
In the rat heart, Kcnk2 mRNA and protein expression has been described in both atrial and ventricular tissue samples (Table 2) [28,29,32,33,149]. However, in the mouse heart, most studies describe ventricular-dominant K 2P 2.1 (TREK-1) expression or mRNA abundance patterns [16,26,41]. Abundant K 2P 2.1 (TREK-1) expression was also detected in the porcine heart, with the highest expression levels in the sinoatrial and atrioventricular nodal tissue [36,37] and in human cardiac tissue samples, where again ventricular dominant K 2P 2.1 (TREK-1) expression could be observed [10,37,40,41]. Interestingly, a transmural gradient of ventricular K 2P 2.1 (TREK-1) expression levels was described with endocardial expression levels 17-fold higher than that in the epicardium, [30,149]. Strikingly, this gradient seems to parallel transmural changes in stretch-activated potassium currents, as mechanical stretch has been shown to cause increased action potential shortening in subendocardial cardiomyocytes compared to the subepicardium [150]. In a similar fashion chloroform-activated K 2P 2.1 (TREK-1)-like currents are significantly larger in endocardial than epicardial cells [30].
Homodimeric K 2P 2.1 (TREK-1) channels are inhibited by the anticonvulsant drugs valproate, gabapentin and carbamazepine [102] by the antidepressants like fluoxetine, paroxetine, citalopram or escitalopram (Table 3) [96,102], and the antipsychotics haloperidol or clozapine [101]. While some of these interactions would only be relevant at supratherapeutic plasma levels, others already have an impact in the physiological range [141]. It has therefore been speculated whether the blockade of cardiac K 2P 2.1 (TREK-1) channels could contribute to the proarrhythmic potential of these compounds [41,141]. Remarkably, K 2P 2.1 (TREK-1) knockout was shown to cause a phenotype of QT interval prolongation, linking loss of cardiac K 2P 2.1 (TREK-1) to QT prolongation [151]. Likewise, antiarrhythmic drugs were described to block K 2P 2.1 (TREK-1) channels: Vaughan Williams class I compounds lidocaine, mexiletine and propafenone, class III antiarrhythmic drugs dronedarone and vernakalant, the beta-blocker carvedilol and late sodium current inhibitor ranolazine were identified as in vitro K 2P 2.1 (TREK-1) inhibitors (Table 3) [43,82,84,104,106,109]. Since IC 50 levels are mostly in the supratherapeutic range, it is unclear to what extent inhibition of K 2P 2.1 (TREK-1) contributes to the antiarrhythmic effects of these compounds.
In isolated rat ventricular cardiomyocytes the mechano-, pH-, and arachidonic acidsensitive potassium current I KAA displays a number of further features like activation by volatile anesthetics, inhibition by cAMP analogues as well as beta-adrenergic receptor agonists, the absence of a relevant voltage dependency, a specific single-channel conductance and burst mode activity, which identify it as a K 2P 2.1 (TREK-1) current (Table 4) [7,28,29,32,33,149,152]. Further, resting membrane potentials of chicken embryo-derived atrial cardiomyocytes are regulated by K 2P 2.1 (TREK-1) [153]. Finally, cardiomyocyte-specific K 2P 2.1 (TREK-1) knockout mice exhibit a phenotype of stress-induced sick sinus syndrome and prolongation of QT intervals that could be reproduced in a transgenic model which employed C-terminal truncation of beta IV spectrin to disrupt its interaction with K 2P 2.1 (TREK-1), thereby impairing intracellular K 2P 2.1 (TREK-1) protein trafficking [27,151]. In a similar fashion, knockout of K 2P 2.1 (TREK-1) channel surface targeting by its protein partners POPDC1 or POPDC2 revealed a phenotype of exercise-induced and age-dependent sick sinus syndrome [154,155], while a double-knockout mouse displayed AV conduction disturbance [156]. Moreover, a familial autosomal recessive POPDC1 mutation has been associated with the phenotype of limb-girdle muscular dystrophy type X2 in combination with AV block [157] and POPDC2 mutations have been shown to cause AV block without a skeletal muscle phenotype [158].The fact that K 2P 2.1 (TREK-1) channels are activated in acidosis and by mechanical stress has given rise to speculation about a role of this channel in the development of cardiac arrhythmias for more than two decades [28]. Metabolic changes associated with myocardial ischemia lead to a decrease in pH. By activating K 2P 2.1 (TREK-1), this can cause a dispersion of repolarization and consecutively the development of arrhythmias. Similarly, altered wall tension due to hypertension, valvular vitiation, in the margins of myocardial scars, or AF may activate K 2P 2.1 (TREK-1) [141,158,159]. Recently, a heterozygous missense mutation (I267T) of K 2P 2.1 (TREK-1) was identified in a patient with idiopathic right ventricular outflow tract tachycardia [160]. This mutation results in an amino acid exchange from isoleucine to threonine in close proximity to the selectivity filter of the channel, leading to increased stretch sensitivity and sodium permeability.      Human Patient-derived tissue samples, iPSC mRNA (RT-qPCR) Downregulation of ventricular mRNA levels in non-ischemic heart failure iPSC: KCNK17 knockdown led to APD prolongation [22] Human, Mouse Index patient HL-1 cells (cultured cardiomyocyte cell line), mRNA (RT-qPCR) A patient suffering from progressive and severe cardiac conduction disorder in combination with idiopathic ventricular fibrillation was identified to carry both, a splice site mutation in the sodium channel gene SCN5A as well as a gain-of-function mutation in the KCNK17 gene HL-1 cells: KCNK17 knockdown overexpression led to APD shortening [5] Human In a murine model of transverse aortic constriction (TAC)-induced pressure overload upregulation of ventricular Kcnk2 mRNA expression was described [16]. In a similar fashion, K 2P 2.1 (TREK-1) protein levels were increased in a rat model of isoproterenolinduced left ventricular hypertrophy [149]. Global K 2P 2.1 (TREK-1) knockout mice showed an exaggerated form of pressure overload-induced concentric ventricular hypertrophy, which could be prohibited only by fibroblast-specific deletion of K 2P 2.1, (TREK-1) whereas the cardiomyocyte-specific knockout of K 2P 2.1 (TREK-1) resulted in cardiac dysfunction under pressure-overload conditions [161]. In a murine atrial fibrillation (AF) model of CREM-Ib∆C-X transgenic mice, downregulation of atrial K 2P 2.1 (TREK-1) mRNA and protein levels were observed [16,41]. It, however, remains uncertain whether this is also the case for AF patients: while one study described AF-associated downregulation of atrial K 2P 2.1 (TREK-1) [37] others merely describe a trend that does not reach statistical significance [10,40,41]. One possible explanation is the remote regulation of atrial K 2P 2.1 (TREK-1) expression by ventricular heart failure, a mechanism recently described for K 2P 3.1 (TASK-1) [40] and also observed for K 2P 2.1 (TREK-1) in another study [41]. Indeed, in contrast to the other study, the cohort of patients characterized in the former study was performed in patients who all suffered from severe heart failure. In a similar fashion, a strong trend towards downregulation of atrial Kcnk2 mRNA could be observed in a murine model of TAC-induced pressure overload [16]. Furthermore, downregulation of atrial K 2P 2.1 (TREK-1) protein expression was described in a porcine model of combined AF and heart failure [36] and gene therapeutic restoraton of K 2P 2.1 (TREK-1) expression was able to attenuate the AF phenotype [37].
For a more detailed description of the cardiac role of K 2P 2.1 (TREK-1), we would like to refer to the following literature [41,141,158].

K 2P 3.1 (TASK-1)
Among the entire K 2P family, K 2P 3.1 (TASK-1) is the channel with the best characterized cardiac significance. K 2P 3.1 (TASK-1) channels are expressed in neuronal tissue, cardiomyocytes, vascular smooth muscle cells, the carotid body glomus, the adrenal gland, brown adipose tissue and immunocytes, where they control important physiological processes [2,115]. K 2P 3.1 (TASK-1) channels are regulated by a number of different stimuli, such as pH level, hypoxia, PKA, PKC, or PLC activity, and several drugs like volatile anesthetics [2].
Several clinically relevant antiarrhythmic drugs have been identified to inhibit homodimeric K 2P 3.1 (TASK-1) channels at either physiological or subtherapeutic concentrations (Table 3). Among them are the class I antiarrhythmic drugs propafenone, mexiletine, lidocaine, and quinidine [104,122,123], the betablockers propranolol and carvedilol [42], class III antiarrhythmics amiodarone and dronedarone [82,110] as well as cardiac glycosides [111] and ranolazine [109]. The respiratory stimulant doxapram was further identified as a potent blocker of both K 2P 3.1 (TASK-1) and K 2P 9.1 (TASK-3) channels through which it presumably exerts the main part of its respiratory drive-increasing effect [119,175]. Furthermore, preclinical experimental antiarrhythmic drugs developed as specific inhibitors of the K V 1.5 channel (A239 [AVE1231], A1899 [S20591], AVE0118, S9947, MSD-D, and ICAGEN-4) are potent K 2P 3.1 (TASK-1) inhibitors [117]. Although no direct structural similarities of the pore regions of both channels could be detected, these compounds were shown to be 1.4to 70-fold more potent K 2P 3.1 (TASK-1) inhibitors as compared to K V 1.5 [117]. In addition, bisamides represent a new class of high-affinity K 2P 3.1 (TASK-1) inhibitors with IC 50 values in the single-digit nanomolar range, as in the case of compound ML365 (Table 3) [116].
Availability of high-affinity inhibitors enables functional detection of K 2P 3.1 (TASK-1) currents in isolated cardiomyocytes. K 2P 3.1 (TASK-1) currents were isolated from rat ventricular cardiomyocytes by lowering pH, activation of cardiac α1-adrenergic receptors and by administration of the inhibitor A293 (Table 4) [15,162,163]. Patch-clamp measurements of murine K 2P 3.1 (TASK-1) current could be confirmed by the use of Kcnk3 knockout mice [25] and likewise, functional detection of K 2P 3.1 (TASK-1) currents was achieved by patch-clamp technique in isolated porcine [52][53][54]164] and human atrial cardiomyocytes, where a significant APD prolongation could be demonstrated [10,39,40,53,56]. Under physiological conditions, I TASK-1 was identified to carry up to 28% of the background potassium current in isolated human atrial cardiomyocytes [39].
In induced pluripotent stem cell-(iPS-) derived cardiomyocytes (iPSC), APD values could be prolonged by transfection of K 2P 3.1 (TASK-1) siRNA [22]. In a zebrafish model, a decreased heart rate was observed after K 2P 3.1 (TASK-1) knockdown, which was accompanied by an increased atrial diameter [165]. In excised guinea pig hearts, APD remained unchanged upon K 2P 3.1 (TASK-1) inhibition with A293 or ML365. Switching the pH level from pH 7.4 to 7.8, however, resulted in significant prolongation of atrial effective refractory periods [49]. Global Kcnk3 knockout mice exhibited a phenotype of QTc prolongation (around 30%), prolongation of single cell APDs or monophasic action potentials and a broad QRS complex [25,26]. In transgenic Kcnk3 knockout rats, APD prolongation as well as resting membrane depolarization was described [163].
In a porcine large animal model of AF, atrial K 2P 3.1 (TASK-1) expression was found to be significantly upregulated (TaqMan qPCR, western blot, patch-clamp electrophysiology) [52,141,164]. These results could also be reproduced on atrial tissue samples from atrial fibrillation patients (TaqMan qPCR, microarray, bulk RNAseq, western blot, patchclamp electrophysiology) [10,41,55,57]. Considering its atrial-specific expression, its effect on atrial APD, and its upregulation in patients with AF, K 2P 3.1 (TASK-1) channels combine several properties that make it an ideal molecular target for the treatment of AF.
Inhibition of K 2P 3.1 (TASK-1) in cardiomyocytes from AF patients has been shown to counteract AF-induced APD shortening [104,154]. After administration of A293 (200 nM), APDs of atrial cardiomyocytes isolated from AF patients could be prolonged around 30% to values observed in sinus rhythm controls [104,154]. After intravenous application of K 2P 3.1 (TASK-1) inhibitors in healthy control pigs, significant prolongation of both, atrial effective refractory periods and ADP values pointed towards class III antiarrhythmic effects of K 2P 3.1 (TASK-1) inhibition [53,54]. Furthermore, the inducibility of atrial arrhythmias was significantly reduced by K 2P 3.1 (TASK-1) inhibitors in different studies [176][177][178]. In a similar fashion, intravenous administration of K 2P 3.1 (TASK-1) inhibitors A293 and doxapram led to rapid, safe and successful cardioversion of artificially induced AF episodes in a porcine large animal model [53,54]. These antiarrhythmic effects could further be employed for rhythm control in a porcine model of burst pacing induced "persistent" AF, induced via implanted pacemakers using a biofeedback algorithm [53,164] and reproduced with an AAV-mediated anti-K 2P 3.1 (TASK-1) gene therapy approach [52]. Based on these encouraging results, the currently ongoing DOCTOS trial (doxapram conversion to sinus rhythm; EudraCT No: 2018-002979-17) was started, which investigates whether the FDA and EMA approved K 2P 3.1 (TASK-1) inhibitor doxapram can cardiovert AF in patients [2,179].
Interestingly, also reduction of atrial K 2P 3.1 (TASK-1) expression was linked to AF as in a dog model of postoperative AF, a phosphorylation dependent downregulation of K 2P 3.1 (TASK-1) was reported [50] and CREM-TG AF mice display atrial downregulation of K 2P 3.1 (TASK-1) in conjunction with massive atrial dilatation and scarring [16]. Patients who suffer from reduced left ventricular ejection fraction display reduced atrial K 2P 3.1 (TASK-1) expression, independently from their rhythm state [40]. Finally, three genetic variants (two kozak variants and missense variant K 2P 3.1 (TASK-1) V123L mutation all of which reduce the expression or channel function) were found in patients with familial AF [49].
In addition to its role in the control of heart rhythm, K 2P 3.1 (TASK-1) is also discussed as a regulator of cardiac energetics and metabolic function, as Kcnk3 knockout mice were protected from pressure overload-induced cardiomyopathy. Compared to wild-type littermates, Kcnk3 knockout mice showed a preservation of systolic as well as diastolic function and a relative abrogation in concentric left ventricular hypertrophy upon TAC-induced pressure overload [46]. Moreover, K 2P 3.1 (TASK-1) channels were described to be expressed in in human pulmonary artery smooth muscle cells, where they serve as regulators of the basal membrane potential and consecutively regulate pulmonary vascular tone [180]. Furthermore, KCNK3 loss-of-function mutations were found to cause idiopathic pulmonary arterial hypertension [166] and acute pharmacological K 2P 3.1 (TASK-1) inhibition in pigs led to a mild but significant increase in invasively measured pulmonary arterial pressure [164]. In the context of adrenal KCNK3 expression, a role of the K 2P 3.1 (TASK-1) channel in aldosterone secretion and blood pressure control is further discussed. Global Kcnk3 knockout mice display a phenotype of mild hyperaldosteronism [181] and single nucleotide polymorphisms in the KCNK3 gene were associated with plasma aldosterone levels [182]. Accordingly, elevated systolic blood pressure values were described in the Kcnk3 knockout mouse [25]. Finally, K 2P 3.1 (TASK-1) channels are also discussed to be involved in regulating function of immune cells and in thermogenesis in brown adipose tissue [183]. Thus, there is a need for further studies that exclude systemic side effects in the use of TASK-1 inhibitors for treatment of AF.

K 2P 4.1 (TRAAK)
Although it was suspected about 20 years ago, that the K 2P 4.1 (TRAAK) channel, based on northern blot analysis, might be mainly expressed in the human heart there is little evidence to date for a cardiac role of this K 2P channel. Several studies reported cardiac KCNK4 mRNA expression, mostly with atrial predominant expression patterns (TaqMan qPCR; Table 2) in human as well as in murine heart tissue samples [10,22,26,41]. Compared with other cardiac ion channels, however, expression levels were relatively low [10,16,41]. A mild inhibitory effect of vernakalant and the late sodium channel blocker ranolazine has also been described for hK 2P 4.1 (TRAAK) homodimeric channels (Table 3) [83,109].
Kcnk4 knockout mice were reported to display smaller ischemic areas upon cerebral infarction. No obvious phenotype of heart rhythm disorder or heart failure was described, and the mice were reported as viably and healthy [167,168]. We are, however, not aware of any studies that explicitly study the cardiac phenotype of these transgenic mice (Table 4).

K 2P 5.1 (TASK-2)
Shortly after the first description of the KCNK5 gene, RT-PCR experiments had already indicated robust cardiac abundance of KCNK5 mRNA [184], while other studies (RT-PCR) considered the cardiac mRNA levels to be rather low (Table 2) [22,23,26,38]. Our own studies indicated atrial predominant KCNK5 mRNA abundance within the human and murine heart [10,16]. Further, a trend towards downregulation of atrial KCNK5 mRNA in patients, suffering from chronic AF was noted that did not reach statistical significance [10]. K 2P 5.1 (TASK-2) homodimers are a molecular target on volatile and amide type local anesthetics (Table 3) [185,186] and inhibited by supratherapeutic concentrations of ranolazine [109]. siRNA transfection experiments pointed towards a functional role of K 2P 5.1 (TASK-2) in setting the membrane potential of pulmonary artery myocytes [187]. In the diabetic rat model with sinus bradycardia, mentioned above, downregulation of cardiac Kcnk5 mRNA expression was reported (Table 4) [19]. Finally, genome-wide association studies could identify a risk locus, associated with the development of coronary artery disease and migraine within the KCNK5 gene [188].
Breeding of global Kcnk5 knockout mice resulted in a small number of female homozygous offspring, pointing towards a phenotype which might cause antenatal mortality [169]. Further, Gerstin et al. reported that one homozygote female animal was found dead in the cage at 12 days of age [169]. However, whether this was associated with cardiomyopathy or arrhythmia remains speculative.

K 2P 6.1 (TWIK-2)
Robust cardiac expressions patterns of KCNK6 mRNA, derived from RT-PCR were described [10,18,22], while others report mild to moderate cardiac expression of this channel (RT-PCR, WB; Table 2) [15,23,26]. Interestingly, mRNA levels were reported to be significantly higher in the adult as compared to the neonatal rat heart [18]. Furthermore, abundant Kcnk6 mRNA levels were found in rat saphenous arteries [189]. Upon TACinduced pressure overload, an upregulation of murine ventricular Kcnk6 mRNA could be observed (Table 4) [16]. Kcnk6 deficient mice are hypertensive and display elevated RV pressure level as well as enhanced vascular contractility which was linked to enhanced rho kinase activity [170][171][172]. The physiological relevance of K 2P 6.1 (TWIK-2) is under debate because these channels conduct only low currents in the heterologous expression system [82]. It further was recently reported that K 2P 6.1 (TWIK-2) channel subunits give rise to functional K 2P currents in endolysosomes, where they affect the size and number of lysosomes [190] so it remains unclear whether the cell membrane is indeed the actual site of action of these channels.

K 2P 7.1 (TWIK-3)
The mainly neuronally detected K 2P 7.1 (TWIK-3) channel is a silent K 2P channel without proven potassium conductance in heterologous expression systems [191]. Only very low cardiac expression levels have been described for KCNK7 (RT-PCR, TaqMan qPCR; Table 2) [10,23]. It was, however speculated whether its mRNA expression might be upregulated in atrial tissue samples, derived from AF patients [63]. Although not explicitly cardiac characterized, a global Kcnk7 knockout mouse showed no obvious cardiac phenotype. Homozygous transgenic mice and wild-type littermates did not differ significantly in general appearance, gross anatomy, locomotion, or overt behavior (Table 4) [173].
Echocardiographic characterization of Kcnk9 knockout mice revealed a phenotype of concentric left ventricular hypertrophy with preserved ejection fraction (Table 4) [46]. In contrast to Kcnk3 knockout mice, however, these animals are not TAC resistant, and heart failure symptoms are more likely to occur at a later time point [46]. Downregulation of ventricular KCNK9 mRNA expression (TaqMan qPCR) in heart failure patients might point towards a pathophysiological role of this channel [22].
Single channel patch-clamp measurements, performed in isolated human atrial cardiomyocytes were able to detect a channel with characteristics corresponding to a heteromer of K 2P 3.1 (TASK-1) and K 2P 9.1 (TASK-3) [56]. However, besides this heteromeric and homodimeric K 2P 3.1 (TASK-1) channels, no current corresponding to a homodimeric K 2P 9.1 (TASK-3) channels could be detected. Functional studies in motoneurons or in rat carotid body glomus cells indicate that the K 2P 3.1 (TASK-1)/ K 2P 9.1 (TASK-3) heterodimer portion was about 52-75% and thus only a minority of K 2P 3.1 (TASK-1) channels are expressed as monomer at the cell surface [192,193]. Since the pharmacological properties of homodimeric and heterodimeric channels differ, heterodimerization has to be taken into account when targeting the K 2P 3.1 (TASK-1) channel in the treatment of cardiac arrhythmias.
A rare genetic disease, KCNK9 imprinting syndrome, also known as Birk-Barel Syndrome is inherited in an autosomal dominant, maternally imprinted manner and associated with congenital central hypotonia, severe feeding difficulties, delayed development, and dysmorphic manifestations [194]. While no direct cardiac manifestation has been described to date, affected individuals may develop obstructive sleep apnea syndrome, which is particularly interesting because it again links the K 2P channels of the TASK subfamily to this disease entity.
Hopefully, the recently available high-affinity K 2P 9.1 (TASK-3) inhibitors and activators will help to answer the question of the functional relevance of K 2P 9.1 (TASK-3) channels in cardiomyocytes.

K 2P 10.1 (TREK-2)
The role of K 2P 10.1 (TREK-2) channel subunits has so far been characterized mainly in the central nervous system (CNS), where this channel shows ubiquitous expression. However, a KCNK10 knockout mouse showed remarkably few neurobehavioral phenotypes besides discrete abnormalities in anxiety-related behavior [174]. A cardiac phenotype of this mouse has not been described yet. Pharmacological in vitro measurements revealed vernakalant and carvedilol as inhibitors of K 2P 10.1 (TREK-2) homodimer channels (Table 3) [43,83]. Low cardiac mRNA abundance was described by our group and others (RT-PCR, TaqMan qPCR; Table 2) [10,15,22,40]. However, the expression patterns appeared atrial-predominant both in murine and patient-derived samples [10,41]. No relevant changes of K 2P 10.1 (TREK-2) expression could be detected in murine disease models of TAC-induced pressure overload or CREM-TG AF (Table 4) [16]. However, in right and left atrial patient-derived tissue samples, significant mRNA upregulation was demonstrated upon systolic heart failure [41].
The observation of reduced KCNK13 mRNA levels in patients with chronic AF or heart failure, which could also be recapitulated in a porcine large animal model of combined AF and heart failure might point towards a physiological role of K 2P 13.1 (THIK-1) currents in regulating atrial electrophysiology [10,40,129]. Finally, ventricular expression levels of KCNK13 mRNA, were described as unchanged in heart failure patients (Table 4) [22].
Reports of reduced KCNK17 mRNA levels in atrial fibrillation [10] and heart failure [22,40] suggest a role for K 2P 17.1 (TALK-2) in the pathophysiology of important cardiac pathologies. K 2P 17.1 (TALK-2) channel subunits were described to heterodimerize with atrial K 2P 3.1 (TASK-1), thereby modulating biophysical and pharmacological properties of atrial I TASK-1 [198]. In heterologous expressions systems, K 2P 17.1 (TALK-2) channel homodimers were reported to be activated by propafenone, quinidine, mexiletine, propranolol, vernakalant, and metoprolol [75]. Amiodarone, sotalol, verapamil, and ranolazine were further described to inhibit K 2P 17.1 (TALK-2) homodimers (Table 3) [75,83]. In iPSC, suppression of K 2P 17.1 (TALK-2) expression was shown to prolong APD (Table 4) [22] while overexpression of K 2P 17.1 (TALK-2) shortened APD levels in the cultured, cardiomyocyte derived HL-1 cell line [5]. Recently, a patient suffering from progressive and severe cardiac conduction disorder in combination with idiopathic ventricular fibrillation was identified to carry both, a splice site mutation in the sodium channel gene SCN5A as well as a mutation in the KCNK17 gene [5]. This K 2P 17.1 (TALK-2) G88R mutation, located in the first extracellular pore loop was shown to increase K 2P 17.1 (TALK-2) currents to about three times upon heterologous expression. Overexpression of K 2P 17.1 (TALK-2) G88R in spontaneously beating HL-1 cells was shown to result in a reduction of the beating frequency, hyperpolarization of the membrane potential and a strong slowing of the upstroke velocity [5].
Single nucleotide polymorphisms in the KCNK17 gene which increase K 2P 17.1 (TALK-2) channel subunit expression levels are associated with the occurrence of ischemic stroke in Caucasians but not in a Chinese population [137,199]. This observation links the channel once again to the pathophysiology of atrial fibrillation. KCNK17 was further proposed as a genetic modifier of long QT syndrome type 2 severity, as a common KCNK17 gain-offunction variant was shown to be LQTS protective by promoting APD shortening [74].
The cardiac characterization of the K 2P 17.1 (TALK-2) channel is complicated by the fact that to date no specific inhibitors are available that would allow functional studies (Table 3). Furthermore, no ortholog to the KCNK17 gene could be identified in mice and the porcine K 2P 17.1 (TALK-2) channel subunit does not appear to show functional activity after heterologous expression in Xenopus laevis oocytes (unpublished observation of our lab).

K 2P 18.1 (TRESK)
KCNK18 mRNA, encoding K 2P 18.1 (TRESK) channel subunits was detected in human spinal cord, trigeminal ganglia, and brain but not in the heart (RT-PCR and TaqMan qPCR; Table 2) [10,61,77,78]. Accordingly, K 2P 18.1 (TRESK) channels are supposed to play a key role in pain perception and KCNK18 was identified as a potential susceptibility gene for migraine, while a cardiac role of this channel is rather unlikely [1]. TRESK channels may nevertheless exert indirect effects on the cardiovascular system: For example, high-fat dietinduced vagal afferent dysfunction has been described to be mediated via upregulation of K 2P 18.1 (TRESK) [200]. Heterologously expressed K 2P 18.1 (TRESK) channel homodimers are inhibited by lidocaine, verapamil, quinidine and apamin (Table 3) [76,200].

Conclusions
Overall, K 2P channels are an exciting and relevant new potassium channel class with relevance to a wide variety of disease conditions. For several members, reproducible mRNA regulation patterns in atrial fibrillation, heart failure and other cardiac disease could be described. However, the functional consequence remains difficult to assess, especially in cases where no specific channel inhibitors are available (Table 3), since surface expression and current amplitude in cardiomyocytes cannot be directly inferred from mRNA expression [11]. Further, the actual significance of the individual K 2P subgroups, some of which show only weak expression patterns, merits further investigation. To date, little is also known about the differential expression of K 2P channels in different cardiac cell populations and the consequence of remodelling in different cell types. In this regard, single cell next generation sequencing technology is expected to provide further evidence soon. Furthermore, computational models of cardiac electrophysiology must consider effects of K 2P channels. Taken together, emerging evidence suggests that K 2P channels play an important role in cardiac repolarization and in the development of various cardiac arrhythmias such as atrial fibrillation, conduction disorders, and ventricular proarrhythmia that goes far beyond the role of unspecific leak currents.
Author Contributions: Conceptualization, F.W., N.F. and C.S.; writing-original draft preparation, F.W.; writing-review and editing, C.S. and N.F.; visualization, F.W.; supervision, N.F.; project administration, C.S. All authors have read and agreed to the published version of the manuscript.
Funding: This research received no external funding.