KCNE1 and KCNE3: the yin and yang of voltage-gated K⁺ channel regulation

Abstract

The human KCNE gene family comprises five genes encoding single transmembrane-spanning ion channel regulatory subunits. The primary function of KCNE genes appears to be regulation of voltage-gated potassium (Kv) channels, and the best-understood KCNE complexes are with the KCNQ1 Kv α subunit. Here, we review the often opposite effects of KCNE1 and KCNE3 on Kv channel biology, with an emphasis on regulation of KCNQ1. Slow-activating I_Ks channel complexes formed by KCNQ1 and KCNE1 are essential for human ventricular myocyte repolarization, while constitutively active KCNQ1-KCNE3 channels are important in the intestine. Inherited sequence variants in human KCNE1 and KCNE3 cause cardiac arrhythmias but by different mechanisms, and each is important for hearing in unique ways. Because of their contrasting effects on KCNQ1 function, KCNE1 and KCNE3 have proved an invaluable tool in the mechanistic understanding of how channel gating can be manipulated, and each may also provide a window into novel insights and new therapeutic opportunities in K⁺ channel pharmacology. Finally, findings from studies of Kcne1^−/− and Kcne3^−/− mouse lines serve to illustrate the complexity of KCNE biology and KCNE-linked disease states.

Graphical abstract

Introduction

Bioelectricity - the electrical activity generated by living organisms - is observable across all six kingdoms of life, and essential in higher animals for processes including fertilization, neuronal firing, and muscular contraction. Electrical signals are conducted primarily by ion channels, which have evolved the capability to selectively and rapidly conduct charged, aqueous ions across the hydrophobic barrier established by the plasma membrane.

Within the ion channels, the S4 superfamily contains those that respond to changes in membrane potential via a voltage sensor that includes the S4 domain. The S4 is so named because it is the fourth transmembrane (TM) segment in the 6-TM module that constitutes the pore-forming (α) subunit of eukaryotic voltage-gated potassium (Kv) channels and some prokaryotic voltage-gated sodium (Nav) and calcium (Cav) channels¹. While each of these voltage-gated channel classes achieve their respective ion selectivity by distinct mechanisms, the atomic structure and signature sequence for K⁺ selectivity is highly conserved between primitive tetrameric bacterial K⁺ channels lacking a voltage sensor², tetrameric mammalian Kv channels carrying the S4³, and even dimeric “two-pore” K2P channels such as human K2P1⁴. Mammalian Nav and Cav channel α subunits consist of four repeats of the 6-TM module whereas mammalian Kv channels form as a tetramer of four non-covalently bound 6-TM α subunits (Figure 1A).

Figure 1. The KCNE gene family.

A. Left, Topology diagram of a 6TM spanning Kv α subunit with voltage sensing domain (VSD) and pore domain highlighted. Right, 1TM topology of a KCNE subunit. Ext, extracellular; Int, intracellular.

B. 4:2 α:β stoichiometry of a KCNQ1-KCNE1 complex.

C. Sequence alignment of the KCNE gene family. Transmembrane (TM) domain boxed; highly conserved residues/features indicated with gray shading.

In Kvs, Navs and Cavs, the voltage sensor is attached to the channel “gate” and when the voltage sensor senses a change in membrane potential and moves in response to this, the movement is communicated to the gate to open or close the pore depending on the channel type; all three channel types are opened by membrane depolarization¹. Hyperpolarization-activated, cyclic nucleotide-gated (HCN) channels are activated (opened) by membrane hyperpolarization; these channels pass both Na⁺ and K⁺ and are referred to as pacemaker channels because they trigger subsequent action potentials in response to undershoot from the previous action potential, and are essential for controlling the heart beat and some rhythmic neuronal firing⁵.

Of all the known Kv channel α subunits, only one, KCNQ1, has thus far been discovered to have the ability to ostensibly lose its voltage dependence. KCNQ1 co-assembles with members of the KCNE family of single transmembrane segment K⁺ ion channel ancillary or β subunits (Figure 1A–C) to enable it to juggle such contrasting roles as cardiac myocyte repolarization, and epithelial cell ion homeostasis/transporter regulation⁶. KCNQ1 exists as a tetramer of α subunits in complexes with two KCNE1 subunits at the plasma membrane^7–9, although stoichiometric variability (between 4:2 and 4:4, α:β) has also been proposed ^10,11 (Figure 1B). The stoichiometry of other Kv-KCNE complexes other than KCNQ1-KCNE1 has not been reported. Channels formed by KCNQ1 and KCNE1 - originally named slowly activating K⁺ current (IsK)¹² and also referred to as Minimal K⁺ channel (MinK) ¹³ - open very slowly, and less easily at negative voltages than homomeric KCNQ1 channels^14,15 (Figure 2). In sharp contrast, KCNE3 (originally named MinK-related peptide 2, MiRP2) ¹⁶ has even more striking effects, locking open the KCNQ1 S4 to almost entirely remove the voltage dependence of KCNQ1 activation, converting it to a K⁺ leak channel^17,18. This review covers the mechanisms and physiological roles of KCNE1 and KCNE3 regulation of KCNQ1 and other Kv channels. A recent companion review was focused on KCNE2¹⁹.

Figure 2. Functional effects of KCNE1 and KCNE3 on KCNQ1.

Example traces recorded by two-electrode voltage clamp from oocytes injected with cRNA encoding KCNQ1 alone or with KCNE1 or KCNE3. Voltage protocol shown at top. Scale bars: vertical, 2 μA; horizontal, 1s. Dotted line indicates zero current level. Traces are from ^{3 and 47}.

KCNE1 and KCNE3 in the context of the KCNE gene family

The first-recognized member of the KCNE family, KCNE1 was discovered by expression cloning via injection of rat kidney mRNA into Xenopus laevis oocytes. One fraction of RNA generated a slowly activating, K⁺-selective and voltage-dependent current measurable by two-electrode voltage-clamp (TEVC) electrophysiology¹². This finding initially caused confusion because the sequenced KCNE1 protein was much smaller than, and bore no sequence similarity to, the known Kv α subunits^20,21. The current generated by KCNE1 strongly resembled the mammalian cardiac Kv current known as I_Ks (slow-activating K⁺ current). Eight years after KCNE1 was cloned, KCNQ1 (then named KvLQT1, also referred to as Kv7.1) was positionally cloned via its linkage to inherited Long QT syndrome (LQTS), a defect in ventricular repolarization that can result from mutations in KCNQ1²². KCNQ1 was found, like KCNE1, to be expressed in human heart. Furthermore, Xenopus laevis KCNQ1 was found to be endogenously expressed in Xenopus oocytes. Co-expression of KCNQ1 and KCNE1 recapitulated the characteristics of I_Ks in cells that, unlike Xenopus oocytes, do not express sufficient endogenous KCNQ1 to yield currents upon heterologous expression of KCNQ1 alone, i.e., Chinese Hamster ovary (CHO) cells. KCNE1 was thus recast as a regulatory, or β, subunit of KCNQ1^14,15, and was swiftly discovered, like KCNQ1, to associate with LQTS^23,24.

KCNE1 was a white elephant for a decade, until the discovery of its relatives using Basic Local Alignment Search Tool (BLAST) screening of online libraries of expressed sequence tags (ESTs)¹⁶. Although the five KCNE genes share relatively little similarity with one another, they each exhibit 1-TM topology with an extracellular N-terminal domain and intracellular C-terminal domain. In addition, each possesses a consensus protein kinase C (PKC) phosphorylation site in the membrane-proximal portion of the cytoplasmic C-terminal region¹⁹. BLAST searching with features of KCNE1 known to be important for its function circumvented the difficulties presented by relatively low family-wide sequence identity, and uncovered KCNE2, 3 and 4 – which we initially termed MinK-related peptides (MiRPs) 1, 2 and 3¹⁶. A fifth member, KCNE1L or KCNE5, was also discovered²⁵.

KCNE1 and KCNE3 regulation of KCNQ1

KCNQ1 is a highly versatile Kv α subunit, which is regulated by a number of different types of protein to produce functionally distinct channel complexes⁶. All five of the KCNE proteins regulate KCNQ1 with distinct functional outcomes, and the effects of KCNE1 and KCNE3 are particularly striking. When heterologously co-expressed with KCNQ1 in mammalian cell lines such as Chinese Hamster ovary (CHO) or human embryonic kidney (HEK) cells, KCNE1 right-shifts the voltage dependence of KCNQ1 activation (opening), slows its activation 5–10 fold^14,15, increases its unitary conductance 4-fold (as measured by noise variance analysis)²⁶, slows its deactivation (closing) and eliminates its inactivation²⁷. In sum, this increases KCNQ1 peak currents at positive voltages several-fold (Figure 2).

In contrast, KCNE3 holds open KCNQ1 such that the complex is constitutively active, no longer relying on membrane depolarization to open¹⁷. Instead, current magnitude (and direction) through KCNQ1-KCNE3 is dictated almost exclusively by driving force (Figure 2). Direct structural studies of the KCNE proteins in isolation (not complexed with α subunits) revealed that the TM segment of KCNE1 is α-helical, and that approximately half the KCNE1 extracellular domain adopts an α helical conformation^28–31. Despite relatively low sequence similarity with KCNE1, the secondary structure distributions of the KCNE3 TM and extracellular domains are quite similar to those of KCNE1 ³². A plethora of structure-function studies have been directed toward understanding the molecular bases for these differences, with The McDonald lab discovered a single residue in the TM region (T58 in KCNE1; V72 in KCNE3) that when switched between KCNE1 and KCNE3 could swap their activation properties. Thus, V72T-KCNE3 slows KCNQ1 activation almost as much as KCNE1 does, and eliminates KCNQ1-KCNE3 constitutive activation³³. Conversely, T58V KCNE1 introduces half as much constitutive current as observed for KCNQ1-KCNE3, and increases the KCNQ1-KCNE1 activation rate 3-fold. Furthermore, concomitant introduction of two KCNE3 residues, to yield F57T, T58V-KCNE1, yielded KCNQ1-KCNE1 channels that were only twofold slower activating than KCNQ1-KCNE3 and almost matched its fractional constitutive current; these channels could be closed by holding for 3 s at −120 mV instead of the normal −80 mV³³.

We subsequently found, using tryptophan tolerance scanning of KCNQ1 S6 followed by mutagenesis of the KCNE1 and KCNE3 residues identified by the McDonald lab, that KCNE1 T58 and L59 interact with KCNQ1 F339 and F340, whereas KCNE3 V72 interacts with KCNQ1 S338³⁴; of note, T58 and L59 mutants were among the first described in KCNE1-linked Jervell and Lange-Nielsen syndrome (JLNS), a cardioauditory syndrome comprising LQTS and sensorineural deafness¹³ (see below). It should be noted that the McDonald lab study utilized the CHO cell expression system³³ whereas our study involved Xenopus oocytes. An interesting phenomenon of KCNQ1-KCNE3 channels is that they exhibit an almost perfectly linear current-voltage relationship when expressed in Xenopus laevis oocytes, while in the typical mammalian heterologous expression systems they exhibit some constitutive activation but a much less linear IV curve. The basis for this difference, be it an artifact of the different techniques used to record whole-cell currents in oocytes (two-electrode voltage clamp, TEVC) versus mammalian cells (patch clamp), or a difference in co-factor expression between the systems, has not to the author’s knowledge been reported. While one may assume the mammalian cells are a truer representation of what might occur in vivo in mammalian tissues, the oocyte system more dramatically highlights the capabilities of KCNE3.

Other KCNE1 and KCNE3 residues are also important for maintaining their tight control of KCNQ1 activation, as uncovered in oocyte studies using TEVC. In KCNE1, the C-terminal domain appears to exert a strong influence on KCNQ1 activation, regulating channel assembly, destabilizing the KCNQ1 open state, altering deactivation rate and modulating rate-dependent facilitation of KCNQ1 activity³⁵. The intracellular end of the KCNE1 transmembrane region is proposed to sit on the S5 end of the KCNQ1 S4-S5 linker (a portion of the channel that links the gate to the voltage sensor)³⁰, and protein-protein interactions were detected between cysteines substituted for wild-type resides H363 (in the S6-proximal region of the KCNQ1 C-terminal domain) and H73, S74 and D76 in the KCNE1 C-terminal domain, and between residues H369 with D76³⁶ (Figure 3A). These interactions probably contribute strongly to slowing KCNQ1 activation and, notably, human KCNE1 S74L and D76N mutations are among the first-described, most-studied, and most pathogenic KCNE1 variants¹⁴. The membrane-proximal portion of the extracellular domain of KCNE1 (residues G40 in the activated state and K41 in the resting state) comes close to KCNQ1 S1 (I145, at the extracellular end), as evidenced from substituted cysteine disulfide trapping. Similarly, a cysteine substituted at KCNE1 L42 can be linked to V324C in KCNQ1 S6, with the result of eliminating deactivation. These interactions themselves might not necessarily be highly influential in the gating of wild-type KCNQ1-KCNE1 complexes but they suggest how KCNE1 orientation might change with respect to KCNQ1 during gating^37,38 (Figure 3A).

Figure 3. Mechanisms of KCNE1 and KCNE3 control of KCNQ1 gating and conductance.

A. Topology diagrams of KCNQ1 with KCNE1 (upper) and KCNE3 (lower) showing residues (numbered where specific residues have been identified as particularly important) and domains identified as being influential in control of KCNQ1 gating by the KCNEs. Arrows: red, KCNE1 or 3 control of KCNQ1 activation; yellow, KCNE1 control of deactivation; orange, KCNE1 control of inactivation; pink, KCNE1 control of conductance; purple, inter-domain interactions within KCNQ1 that are affected by KCNE1; dashed purple, KCNQ1 outer vestibule/selectivity filter flexibility; dashed gray, physical proximity and/or interaction without major functional effects. “−” = acidic residue, “+” = basic residue. Citations appear in main text.

B. Cartoon of possible orientations with respect to KCNQ1 (blue), based on evidence summarized in panel A, of KCNE1 (upper) and KCNE3 (lower) at resting membrane potential. S6′ = the S6 of an adjoining KCNQ1 α subunit. KCNEs are depicted as semi-transparent to avoid concealing KCNQ1 features.

In contrast, in KCNE3, the transmembrane domain appears to override the influence of its C-terminal domain³⁹. In addition, The KCNE3 N-terminal region possesses two acidic residues crucial for the robust constitutive current exhibited by KCNQ1-KCNE3 channels when expressed in oocytes. Using double-mutant thermodynamic cycle analysis, we discovered energetic coupling of KCNE3 residues D54 and D55 specifically to residue R237 in the voltage-sensing S4 helix of KCNQ1⁴⁰. We had previously found that the relative charge paucity of KCNQ1 S4 (+3, the lowest of all known eukaryotic voltage-dependent ion channels) was instrumental in the functional flexibility of KCNQ1 in terms of regulation by KCNE subunits¹⁸. Neutralization of R231 within KCNQ1 S4, by replacement with alanine, converts KCNQ1 to a K⁺ leak channel in the absence of KCNEs, and the same is achieved in related α subunit KCNQ4 by a triple mutation that mimics the charge balance of R231A-KCNQ1^18,41. Homomeric R237A-KCNQ1 channels exhibit slightly right-shifted activation voltage dependence compared to homomeric KCNQ1, but the former accumulate in the open state with repetitive membrane depolarizations, deactivating very slowly upon repolarization¹⁸. In KCNQ1-KCNE3 complexes, KCNQ1-R237 appears to engage in electrostatic interactions with KCNE3 D54 and D55 to help stabilize the KCNQ1 S4, and therefore also the pore, in the activated state, even at negative voltages (deactivation can be seen to occur over several seconds when hyperpolarizing to −120 mV) (Figure 3A). Interestingly, KCNE3 D54 partially protects KCNQ1-KCNE3 channels from the inhibitory effects of loss of plasma membrane phospholipid head groups. Normally, these negatively charged phosphocholines stabilize the activated conformation of S4, facilitating voltage-dependent activation. Removal of these head groups via hydrolysis by, for example, sphingomyelinase C, inhibits activity of channels including Kv2.1⁴² and KCNQ1, but KCNE3 (and not KCNE1) helps maintain KCNQ1 activity in the face of sphingomyelinase C because KCNE3-D54 mimics the S4-stabilizing action of membrane phosphocholines⁴⁰.

As striking as the constitutive activation of KCNQ1 endowed by KCNE3, is the laboriously slow activation endowed by KCNE1. As covered above, the KCNE1 C-terminal domain and to some extent the transmembrane domain may be important in slowing KCNQ1 activation, but is this achieved by slowing the voltage sensor, the pore opening, or both? There are at least two camps in terms of this aspect of the mechanistic underpinnings of slow KCNQ1-KCNE1 activation. The Goldstein and Bezanilla labs approached the problem using cut-open oocyte recording, a technique that permits much better voltage clamp and improved time resolution compared to conventional TEVC, facilitating measurement of gating currents – the current generated from movement of the charged voltage sensor itself. The cut-open oocyte recordings were combined with site-directed fluorimetry to also monitor voltage sensor movement by a complementary approach. They discovered that unlike homomeric KCNQ1, during KCNQ1-KCNE1 activation S4 moves too slowly to permit resolution of the gating current; by fluorimetry S4 was found to move 30-fold slower than that of the oft-studied Shaker Kv channel⁴³.

Nakajo and Kubo found, using methanethiosulfonate (MTS) modification of cysteine-substituted KCNQ1 residues, that accessibility of A226C-KCNQ1, a substituted residue near the extracellular side of KCNQ1 S4, was 13-fold slower when KCNE1 was co-expressed compared to KCNQ1 alone. In contrast, when KCNE3 was co-expressed, KCNQ1 A226C accessibility was voltage-independent. This is consistent with the slow S4 movement model for KCNQ1-KCNE1 activation, although it is also consistent with S4 equilibrium rather than speed of movement being modulated by KCNE1⁴⁴. Subsequent work by Nakajo and Kubo also suggested KCNE1 modulates interactions between S4 and S5 to affect activation gating, which could affect S4 movement speed and/or coupling of S4 to the channel gate⁴⁵. The Seebohm lab found that KCNE1 interacts with both S4 and S5/S6 to form what is probably a stable pre-open state that may slow activation⁴⁶. Utilizing voltage clamp fluorimetry of Xenopus oocyte-expressed channels, the Kass lab formulated a dual-effect model for KCNE1 regulation of KCNQ1, in which KCNQ1 can open after activation of just one S4, while KCNQ1-KCNE1 needs all four S4s to shift before the pore opens. This model incorporates both slow S4 movement, and the need for all four S4s to open, as the basis of slow KCNQ1-KCNE1 activation^47,48. The Fedida lab found using single channel analysis that KCNQ1-KCNE1 flits between a variety of subconductance states, behavior that may or may not also exist in faster channels but could be too rapid to resolve in these cases, and may strongly influence opening and closing rates of KCNQ1-KCNE1⁴⁹.

The jury is clearly still out over the specifics, but in general terms it appears KCNE1 interaction with S4 retards KCNQ1-KCNE1 activation, directly and/or or via altered interaction with S4 and S5/S6. Discrepancies between different studies could arise from different preparations, protocols and interpretations. Interaction of KCNE1 with KCNQ1 S4 may also generate the right-shift in voltage dependence of activation induced by KCNE1, possibly by altering the manner in which S4 senses the stabilizing acidic residues in other, nearby KCNQ1 TM domains^46,50,51.

Low mobility of the KCNE1-bound KCNQ1 S4 was also recently suggested to underlie the slow deactivation conferred by KCNE1⁵². In this comprehensive combination of experimental work, and modeling using elastic network analysis, loss of KCNQ1 C-type (slow) inactivation, another change conferred by KCNE1 co-assembly, was assessed in the context of acidic residues that influence slow inactivation of KcsA (the Steptomyces lividans K⁺ channel used by the McKinnon lab to produce the first high-resolution K⁺ channel structure²). It was suggested that KCNE1 limits positional fluctuations in the KCNQ1 outer vestibule and selectivity filter that drive slow inactivation, thereby eliminating or at least dampening the inactivation process. Finally, the increased conduction of KCNQ1 upon KCNE1 co-assembly was attributed in this study to a shift in KCNQ1 residues G272, V310 and T311 such that they move in concert with an entire “dynamic domain” in the pore. This is proposed to increase the scale of pore movement and therefore increase conductance, and is consistent with the results of previous mutagenesis studies⁵². The contrasting mechanisms by which KCNE1 and KCNE3 exert their effects on KCNQ1 gating and conductance, and possible locations of KCNE1 and KCNE3 within complexes with KCNQ1, are summarized in Figure 3A and B.

In addition, KCNE1 modulates KCNQ1 internalization from the plasma membrane. While KCNQ1 channels can be internalized by Nedd4/Nedd4-like- and Rab5-dependent processes with or without KCNE1 co-assembly^53,54, KCNE1 mediates an alternative process, clathrin-mediated endocytosis of KCNQ1⁵⁵. This dynamin-dependent process is enhanced by PKC phosphorylation of KCNE1 at residue S103⁵⁶, explaining previous findings that S102 phosphorylation diminished currents generated by KCNE1 (in complex with endogenous KCNQ1) in Xenopus oocytes^57,58.

KCNE1 and KCNE3 also exhibit contrasting effects on the functional outcomes of KCNQ1 regulation by other factors. Serine 27 in the KCNQ1 N-terminal region can be phosphorylated by protein kinase A (PKA), a physiologically important mechanism for β-adrenergic stimulation of cardiac I_Ks. Although S27 can be phosphorylated regardless of the co-expressed KCNE isoform, the functional effect also requires KCNE1 (or KCNE2) and the cytoplasmic protein yotiao in the complex, despite the PKA site being on KCNQ1 itself. KCNQ1-KCNE3 complexes do not show the same upregulation in response to PKA phosphorylation of KCNQ1 S27⁵⁹. Of note, KCNE1 increases the sensitivity of KCNQ1 channels to phosphatidylinositol 4,5-bisphosphate (PIP₂), a plasma membrane lipid that acts as a co-factor for a variety of ion channels and augments I_Ks. It is not yet known whether KCNE3 has similar effects, but all members of the KCNE family possess similar residues to those important for modifying the manner in which PIP₂ interacts with KCNQ1⁶⁰.

KCNE1 and KCNE3 regulation of other Kv α subunits

Although KCNE1 and KCNE3 regulation of other Kv channels is not generally as well understood as their effects on KCNQ1, it is interesting to note that their regulation of other α subunits also typically results in disparate outcomes (Table 1). The first non-KCNQ1 α subunit found to be regulated by KCNE1 was the human ether-a-go-go related gene product (hERG, also termed Kv11.1 or KCNH2), the channel primarily responsible for driving human ventricular repolarization, especially at resting heart rates. KCNE1 increases hERG current density by a still unresolved mechanism in heterologous expression studies⁶¹, and probably regulates hERG in the heart^62–64. In contrast, KCNE3 strongly inhibits hERG, again by an unresolved mechanism¹⁷, although the same KCNE3 substitution (R83H) that disrupts ability to constitutively open KCNQ1 channels and augment Kv3.4 currents⁶⁵ also blunts the ability of KCNE3 to inhibit hERG⁶⁶. Interestingly, KCNE1 and KCNE3 are reported to form tripartite complexes with Kv12.2 (also termed ELK2 or KCNH3), inhibiting its surface expression and activation in Xenopus oocytes; this may also occur in mouse brain⁶⁷.

Table 1. Summary of main effects of KCNE1 and KCNE3 on Kv α subunits.

See main text for expanded description of effects.

Kv α subunit	Effects of KCNE1	Effects of KCNE3
KCNQ1	Slows activation, strongly increases peak current in oocytes, CHO and HEK cells, and cardiac myocytes.12,13	Instantaneous activation, small increase in peak current in oocytes and CHO cells.17,33
KCNQ4	Mild upregulation in Xenopus oocytes.67	Inhibits in Xenopus oocytes.17,78
KCNQ5	Slows activation, increases peak current by decreasing inward rectification in HEK, oocytes. 77	Inhibits in HEK cells and in Xenopus oocytes. 77
hERG	Doubles current density in CHO cells.61	Strongly inhibits in Xenopus oocytes.8
Kv1.1	No effect in CHO cells.69,70	?
Kv1.1-Kv1.4	Increases current density 2-fold in CHO cells.69,70	?
Kv1.4	Traps in ER/Golgi in CHO cells. 69,70	?
Kv2.1	Slows activation/deactivation in CHO cells.57	Slowed activation/deactivation in CHO cells.76
Kv3.1	Strongly slows activation/deactivation, accelerates inactivation in CHO cells.71	Weakly slows activation/deactivation, accelerates inactivation in CHO cells.71
Kv3.1-Kv3.2	Strongly slows activation/deactivation, accelerates inactivation in CHO cells.71	Weakly slows activation/deactivation, accelerates inactivation in CHO cells.71
Kv3.1-Kv3.4	Attenuates inactivation by “mopping” up homomeric Kv3.4 in CHO cells.69,70	?
Kv3.2	Strongly slows activation/deactivation, accelerates inactivation in CHO cells.71	Weakly slows activation/deactivation, accelerates inactivation in CHO cells.71
Kv3.3	Inhibits, probably by trapping in ER/Golgi in CHO cells.69,70	?
Kv3.4	Traps in ER/Golgi in CHO cells. 69,70	Augments activity, left-shifts voltage dependence in CHO cells.65
Kv4.2	No effect (with α,β staggered injection) in Xenopus oocytes.73	Inhibits in CHO cells and In SGNs.72
Kv4.3	Slows gating, increases current in HEK cells.75	Inhibits in CHO cells.74
Kv4.3-KChIP2	Accelerates activation. Mildly speeds, slows recovery of, and left-shifts steady state inactivation; all in CHO cells.76	Accelerates activation. Strongly speeds, slows recovery of, and left-shifts steady state inactivation; all in CHO cells.76
Kv12.2	Inhibits in Xenopus oocytes.67	Inhibits Xenopus oocytes.67

KCNE1 and KCNE3 both slow Kv2.1 activation and deactivation, possibly increasing the diversity of Kv2.1 currents in heart and brain^64,68, and one of the rare instances in which these two KCNEs exhibit comparable effects on the activity of a specific Kv channel. In contrast, KCNE1 and KCNE3 have opposite effects on Kv3.4, a rapidly inactivating, N-type α subunit important in skeletal muscle and the CNS – albeit by different mechanisms. KCNE1 robustly inhibits Kv3.4 activity by retaining it early in the secretory pathway (endoplasmic reticulum and/or Golgi apparatus). We discovered that this provides a mechanism by which cells can ensure only mixed N-type/delayed rectifier channel complexes with intermediate inactivation kinetics, and not homomeric N-type α subunit channels, reach the plasma membrane. Both KCNE1 and KCNE2 exhibit this retention ability, and they can retain all three classic N-type, inactivation domain-containing Kv α subunits (Kv1.4, Kv3.3 and Kv3.4) until rescue by same-subfamily delayed rectifiers^69,70. Conversely, KCNE3 augments activity of Kv3.4 channels, by a mechanism involving left-shifted voltage dependence of activation, increased unitary conductance, accelerated recovery from inactivation, and suppression of cumulative inactivation⁶⁵. With respect to regulation of Kv3.1, Kv3.2, and Kv3.1-Kv3.2 heteromers, KCNE1 and KCNE3 both slow activation and deactivation and accelerate inactivation, but the effects of KCNE1 are of much greater magnitude⁷¹.

KCNE3 was recently found to inhibit Kv4.2, and Kv4.2-KCNE3 complexes may contribute to spike frequency and other properties in auditory neurons⁷² (see below), while no effects were observed for KCNE1 with Kv4.2 in Xenopus oocyte co-expression studies (albeit using a staggered, α subunit-first, cRNA injection protocol)⁷³. Further, KCNE3 was found to inhibit Kv4.3 (fourfold reduction in current density) in CHO cells⁷⁴, while in contrast, in HEK cells KCNE1 augmented Kv4.3 current density fivefold and slowed its inactivation ⁷⁵. Association of the cytosolic ancillary subunit, KChIP2, altered these effects. Thus, in tripartite complexes, all KCNEs accelerated activation and inactivation of Kv4.3-KChIP2 channel complexes, although compared to complexes formed with KCNE1, Kv4.3-KChIP2-KCNE3 channels exhibited lower current density, and slightly more negative half-maximal voltage dependence of steady-state activation and inactivation⁷⁶. One or more of the KCNE subunits may thus contribute to functional diversity of cardiac I_to.

Returning to the KCNQ family, in both Xenopus oocytes and HEK cells, KCNE1 reportedly slows the activation of KCNQ5, decreases its inward rectification at highly depolarized voltages, and increases its peak current; conversely, KCNE3 inhibits KCNQ5⁷⁷. Similarly, KCNE1 appears to mildly augment KCNQ4 current when co-expressed in Xenopus oocytes, whereas KCNE3 strongly inhibits KCNQ4 in this expression system^17,78.

KCNE1, KCNE2 and KCNE3 show sex-dependent differences in expression that arise from hormonal influence and determine some sex-specific aspects of mammalian physiology. Cardiac KCNE2 expression is upregulated by estrogen via a direct genomic mechanism⁷⁹. KCNE1 and KCNE3 (but not KCNE2) are expressed in rat colonic crypts and highly enriched in male versus female rodents; during the estrus cycle, colonic expression of KCNE3 (and colonic partner KCNQ1) fluctuates in a manner consistent with KCNE3 downregulation by estrogen, via PKC phosphorylation of KCNE3 serine 82. In heterologous expression experiments, estrogen inhibited KCNQ1-KCNE3 activity, but not that of KCNQ1-KCNE1^80–82.

In contrast, KCNE1, 3 and 5 expression is upregulated in concert with that of the orphan nuclear receptor, estrogen-related receptor γ (ERRγ), in human syncytiotrophoblasts, and is downregulated by hypoxia; KCNE4 displays the reverse pattern. KCNE1 and KCNE3 expression in trophoblast cells is downregulated by shRNA suppression of ERRγ and KCNE1 contains a putative ERRγ response element within its 5′ flanking region⁸³. KCNE1 and KCNE2 expression was also found to be downregulated in ERRγ null mouse hearts, a similar effect was observed for KCNE2 in parietal cells, and KCNE2 (like KCNE1) was found to contain a ERRγ responsive element that binds ERRγ⁸⁴.

Physiological roles for KCNE1: the heart and inner ear

The best-understood physiological roles for KCNE1 are encapsulated by JLNS, an inherited cardioauditory syndrome that results from severe disruption of KCNQ1-KCNE1 channels such as occurs with biallelic mutations in human KCNE1 or KCNQ1^23,85. The cardiac element of JLNS is LQTS, caused by loss of function of ventricular KCNQ1-KCNE1 channels, eroding the repolarization reserve of ventricular myocytes and delaying ventricular repolarization. Thus, the slow- and relatively positive membrane potential-activating KCNQ1-KCNE1 channel, which generates I_Ks, is important for repolarizing human ventricular myocytes to end their action potential in readiness for the next heartbeat^24,86. I_Ks appears to be most important during physical exertion or other contexts when β–adrenergic stimulation increases I_Ks density to compensate for higher heart rates. At other times, I_Kr (generated by hERG, which is regulated in vivo by KCNE1 and KCNE2) is probably dominant in human ventricles^87,88.

KCNE1 mutants more commonly cause the LQT5 form of LQTS, without overt auditory symptoms, and presumably though disruption of I_Ks complex function, although these mutants can also cause loss-of-function of hERG-KCNE1 channels and therefore potentially ventricular I_Kr in vivo⁸⁹. To the author’s knowledge, KCNE1 sequence variation has not be linked to Brugada syndrome or to Short QT syndrome (SQTS), either of which (opposite to LQTS) can result from gain of function in ventricular Kv channels^90,91. However, a KCNQ1 mutation was recently discovered that appears to shorten the QT interval by impairing association of KCNQ1 with KCNE1, leading to accelerated activation and a negative shift in the voltage dependence of activation of I_Ks; the mutation also decreased inactivation of homomeric KCNQ1⁹². Human KCNE1 sequence variants in its extracellular portion are associated with the relatively rare, idiopathic form of atrial fibrillation (AF), with the underlying mechanism considered to be gain of function of atrial myocyte KCNQ1-KCNE1 channels. In addition, a more common KCNE1 polymorphism, S38G, has been linked to altered predisposition to AF and LQTS, with environmental and other genetic factors influencing this predisposition^93–96. The same polymorphism influences predisposition to postoperative atrial fibrillation⁹⁷, a prevalent arrhythmia following cardiothoracic surgery, and KCNE1 is downregulated in the atria of pigs following lung lobectomy⁹⁸. Human KCNE1 sequence variants associated with human disease are summarized in Table 2.

Table 2. KCNE1 gene variants associated or putatively associated with human disease.

AF, Atrial Fibrillation; AV block, atrioventricular block; BrS, Brugada Syndrome; diLQTS, drug induced Long QT Syndrome; diTdP, drug induced torsade de pointes; fs, frame shift; JLNS, Jervell and Lange-Nielsen Syndrome; n.r., not reported; RWS, Romano-Ward Syndrome; Sources include cited papers and ¹⁶². KCNE1 gene accession number: P15382.

Mutation	Disease	In vitro cellular effects
T7I	JLNS85	n.r.
A8V	LQT5, AF	IKs +9 mV positive shift in V1/2 activation. No change in deactivation or activation time constants. IKr loss of function. 134,135
T10M	LQT5135	n.r.
W17X	LQT5135	n.r.
G25V	Lone AF	IKs gain of function 95
S28L	LQT5, AV block (neonatal)	n.r. 135,136
R32H	LQT5	Gating not altered135,137,138
38G versus 38S	AF, AF in elderly patients, postoperative AF, LQT5	IKs loss of function, reduced membrane KV7.1 (KvLQT1) channel93,97,139–144
38G versus 38S	Ménière’s disease (Japanese cohort; not recapitulated in study of Caucasian population)122,123	n.r
V47F	LQT5	hERG Gain of Function145
L51H	LQT5	Does not process properly, not found in membrane145
G52R	LQT5	IKs reduced 50%146
G55S	LQT5135	n.r.
T58P	LQT5135	n.r.
L59P	LQT5135	n.r.
G60D	Lone AF	IKs gain of function; faster deactivation95
R67C	LQT5135	n.r.
R67H	LQT5135	n.r.
K70M	LQT5135	n.r.
K70N	LQT5147	n.r.
S74L D76N	LQT5 (JLNS+RW)24 diLQTS148	D76N: IKs - smaller unitary currents and open probabilities; dominant negative26 IKr – loss of function145 S74L: IKs - loss of function149
N75-fs+34X	LQT5135	n.r.
Y81C	LQT5?	IKs loss of function, Positive shift for voltage of activation150
E83K	LQT5135	n.r.
D85N	LQT5? DiLQTS DiTdP	No effect on IKs V1/2 of activation, or deactivation properties 148,151–159
W87R	LQT5	IKs loss of function, altered gating145
R98W	LQT5?	Disrupts IKs trafficking, right-shifts voltage dependence of activation160
V109I	LQT5	IKs loss of function (36%)161
Q117X	LQT5135	n.r.
T125M	LQT5135	n.r.
P127T	LQT5, RWS137	n.r.

While KCNQ1-KCNE1 channels activate much faster at body temperature than the classic slow activation they exhibit at room temperature⁹⁹ - which makes the effects of KCNE1 particularly apparent - KCNE1 induces other changes in KCNQ1 important for their function together in the heart. As explained above, KCNE1 (together with a raft of cytosolic regulatory proteins) facilitates PKA upregulation of I_Ks in response to β adrenergic stimulation^100–105. KCNE1 also appears to prime KCNQ1 to be ready to respond to the need for elevated heart rate by inducing a reserve of near-open closed states prepared for rapid activation when needed⁸⁷, such that I_Ks complexes can be thought of as the ventricles’ ‘minutemen’. As explained earlier, KCNE1 mediates PKC-stimulated endocytosis of KCNQ1-KCNE1 channels, providing an alternate pathway for downregulation of I_Ks (and also perhaps a therapeutic window for counteracting QT prolongation, if this process can be disrupted)⁵⁶. Finally, elimination of KCNQ1 slow inactivation by KCNE1 may also better enable I_Ks to sustain rapid heart rates without current diminishment.

The non-inactivating status endowed by KCNE1 may also facilitate the established function of KCNQ1-KCNE1 in the inner ear. Although KCNQ1-KCNE1 channels activate voltage-dependently, they are capable of sustained activity at the membrane potentials typical for the apical membrane of strial marginal cells and dark cells of the inner ear (0 to +10 mV). KCNQ1-KCNE1 provides a conduit for K⁺ secretion, thus recycling K⁺ in the endolymph of the inner ear. In the human ear, loss of this recycling is sufficient to cause bilateral sensorineural deafness; Kcnq1 deletion in mice revealed that this is associated with morphological abnormalities because of reduction in the volume of endolymph¹⁰⁶. KCNQ1-KCNE1 channels are, in contrast, not prominent in adult musine ventricular function. Neonatal musine neonatal myocytes exhibit I_Ks and expression of both KCNQ1 and KCNE1 proteins^56,89,107. Adult mouse ventricles express an adrenergic-sensitive steady-state current that is disrupted by Kcnq1 deletion, does not involve KCNE1¹⁰⁸, and could potentially be generated by KCNQ1-KCNE2 complexes. Kcne1 deletion reportedly causes atrial fibrillation in mice, and KCNE1 can be detected in adult mouse atria and conduction system¹⁰⁹, but it is not known which α subunit(s) KCNE1 regulates in these tissues in the mouse – Kv2.1 being one possibility.

Controversies and discrepancies (I). KCNE1 in the brain and kidneys

Typically not considered a neuronal complex, it was recently proposed that KCNQ1-KCNE1 channels may operate in some neurons, and that this could introduce a neural component to cardiac arrhythmias. Thus, mutations in KCNE1 or KCNQ1 might cause seizures that in turn trigger cardiac arrhythmias (perhaps exacerbated by the channels also being dysfunctional in the heart and therefore increasing arrhythmia susceptibility in the face of a trigger)¹¹⁰. Interestingly, Kcnq1^−/− mouse hearts exhibit impaired repolarization in vivo but not when isolated, suggesting Kcnq1-linked LQTS in mice arises from an extracardiac source¹⁰⁶. The controversy continues, centering on whether KCNE1 and KCNQ1 are actually expressed in neurons in mice and people.

Kcne1^−/− mice exhibit altered salt and glucose homeostasis, and reduced NaCl consumption - the latter suggestive of a role for KCNE1 in mouse kidneys or salivary glands¹¹¹. KCNQ1 and KCNE1 are expressed in mouse kidneys but they do not necessarily form a channel there, and the pharmacology and gating kinetics of K⁺ currents in proximal convoluted tubule epithelial cells –where KCNE1 is expressed – are not consistent with KCNQ1-KCNE1 properties. Counterintuitively, the KCNQ1-preferential antagonist, chromanol 293B, had similar functional effects in the proximal tubule to those of Kcne1 deletion, but this may reflect nonspecific action of chromanol 293B at the dosage used (100 μM). The cell-specific expression, biological functions and α subunit partners of KCNE1 in the kidneys remain a matter of debate^112–116. KCNE1 was also previously suggested to form channels with KCNQ1 in the airway epithelium, but it now appears much more likely that KCNQ1-KCNE3 complexes are expressed there instead (see below)¹¹⁷.

Physiological roles for KCNE3: the gut, airway and auditory neurons

KCNE3 is best known for its regulation of KCNQ1 in the intestine. KCNQ1-KCNE3 are located on the basolateral membrane of colonic crypt cells. Their constitutive activation and lack of inactivation enable them to provide a K⁺ recycling conduit that facilitates electrogenic intestinal Cl⁻ secretion. All available evidence for the importance of this process is from the mouse, and specifically the Kcne3^−/− mouse line, and there are currently no reported human disease associations linking KCNE3 (or KCNQ1) mutations to altered intestinal Cl⁻ secretion^17,118. As mentioned above, estrogen downregulates intestinal KCNE3 expression, causing the water retention that occurs in female mice during proestrus. As observed for the intestine, Kcne3 deletion in mice also disrupts tracheal Cl⁻ secretion, and basolateral KCNQ1-KCNE3 complexes are thought to be required for cAMP-stimulated Cl⁻ secretion in the airway epithelium. KCNQ1-KCNE3 channels may also facilitate Na⁺ reabsorption in the airways of mice¹¹⁷. Again, data for the potential role of KCNQ1-KCNE3 in human airway epithelia are currently lacking.

In a different study of Kcne3^−/− mice, a role for KCNE3 in auditory neurons was uncovered. KCNE3 was found to be expressed in spiral ganglion neurons (SGNs), with Kcne3^−/− SGNs utilized as a convincing negative control for expression. Kcne3 deletion age-dependently altered the firing properties and membrane potential variability of SGNs, and resulted in an increase in the transient outward current in these cells. Kv4.2, and not Kv4.3, was found to be expressed in both apical and basal SGNs, where it co-localized with KCNE3. Accordingly, KCNE3 was found to diminish Kv4.2 activity when the two were co-expressed in CHO cells⁷². Therefore, KCNE3 may act as a dampener for Kv4.2 activity in SGNs, permitting dynamic regulation of the transient K⁺ current in these cells without the cell being required to remove Kv4.2 from the membrane, which might be too slow for the cells’ requirements or too energetically costly. The functional effects in SGNs of Kcne3 deletion were only observed during onset of hearing, during the first 12 days. At day 56, only two effects could be observed in SGNs – slightly enhanced afterhyperpolarizations of basal but not apical SGNs, and reduced interspike interval. Thus, KCNE3 might be more important during development of hearing than after auditory maturity, or else compensatory remodeling of other channels might lessen the deleterious consequences of Kcne3 deletion over time⁷².

Controversies and discrepancies (II). KCNE3 in the brain, muscle and auditory system

We previously found expression of KCNE3 transcript by Northern blot in all regions tested of human brain, and by RT-PCR from 12 DIV cultured rat primary hippocampal neurons isolated from embryonic stage 18 Sprague Dawley rats⁶⁸. In addition, we detected KCNE3 protein by western blot in crude membrane preparations isolated from adult Sprague Dawley rat whole brains. We detected KCNE3-Kv2.1 and KCNE3-Kv3.1 complexes in these preparations, using co-immunoprecipitations. Our studies showed that endogenous KCNE3 co-localized with endogenous Kv2.1 but not Kv3.1 in cultured hippocampal neurons, while siRNA knockdown of endogenous KCNE3 in the PC12 cell line upregulated endogenous PC12 cell delayed rectifier K⁺ current. Heterologous co-expression studies in CHO cells were consistent with the functional effects we had observed by knocking down endogenous KCNE3 in PC12 cells, i.e., KCNE3 slowed the activation of and downregulated Kv2.1 current (and downregulated overall activity but reduced inactivation of Kv3.1b)⁶⁸. As mentioned above, KCNE3 was detected by others in auditory neurons, with the added advantage of having a Kcne3^−/− negative control⁷². However, in contrast to our findings, Preston et al reported a failure to detect KCNE3 in mouse brain¹¹⁸.

Similarly, Preston and colleagues reported a lack of KCNE3 expression in mouse skeletal muscle, and a lack of effects of Kcne3 deletion on the skeletal muscle-related parameters they tested in mice: muscle morphology, spontaneous paralysis, myotonia, “movement”, and rotarod performance¹¹⁸. Our studies of KCNE3 in skeletal muscle were driven by the findings of colleagues that the human KCNE3 R83H mutation is associated with periodic paralysis – a skeletal muscle disorder in which sufferers experience bouts of debilitating paralysis. We detected KCNE3 transcript by northern blot in human skeletal muscle, and found KCNE3 protein in rat sartorius muscle by western blot and in the mouse C2C12 myoblast cell line by western blot and immunofluorescence⁶⁵. We determined that KCNE3 co-assembles with and functionally regulates Kv3.4, an N-type Kv channel α subunit known to be expressed in skeletal muscle, and that the R83H mutation in KCNE3, found to associate in individuals with periodic paralysis within two families, impaired function of Kv3.4-KCNE3 channels. Specifically, the mutant disrupted the ability of KCNE3 to left-shift the voltage dependence of Kv3.4 activation, leading to depolarization of resting membrane potential in skeletal myocytes⁶⁵. We subsequently found that the R83H mutation also impairs the ability of KCNE3 to regulate other Kv channels⁶⁶. Further, we found that the histidine introduced in this substitution makes Kv3.4-KCNE3 channels susceptible to block by protons, and could therefore make carriers particularly prone to impaired channel function during rest after exercise, when muscle pH drops drastically. Additionally, we showed that the neighboring residue, S82, is a serine phosphorylation site and that PKC phosphorylation of this residue is required for the left-shift in voltage dependence of Kv3.4 activation endowed by KCNE3¹¹⁹. However, the disease association has been highly controversial, with others finding the R83H sequence variant in unaffected individuals, raising the possibility that it is a benign polymorphism¹²⁰. Another possibility is that carriers need another genetic ‘hit’ or specific environmental conditions to provoke paralysis¹²¹.

Another controversy surrounds the possible association of KCNE3 (and KCNE1) with Ménière’s disease – an inner ear disorder that manifests as tinnitus, periodic hearing loss, and spontaneous episodes of vertigo. Single nucleotide polymorphisms (SNPs) in KCNE1 (112G/A, coding for the well-studied S38G polymorphism) and KCNE3 (198T/C, a synonymous switch within the codon for F66) were found in one study to be enriched in Japanese Ménière’s cohorts compared to unaffected individuals¹²². However, in a study of Caucasians, KCNE1 and KCNE3 SNPs were not found to associate with Ménière’s¹²³. Deep resequencing of KCNE3 in Caucasian people with chronic tinnitus failed to either support or disprove a link, with the authors concluding that more power was required¹²⁴. While both genes are expressed in the auditory system (KCNE3 has been detected in auditory neurons, and all five KCNEs were detected in outer hair cells⁷⁸), and KCNE1 is tightly associated with JLNS^23,85, establishing a firm link between human KCNE3 sequence variation and auditory dysfunction will require further population or linkage studies.

Finally, not so much a controversy or necessarily a discrepancy, KCNE3 has been detected in human heart¹⁰⁴ (although the relative expression levels of KCNEs varies depending on the study^76,125 and may be affected by the sex of the donors, which was not always reported) and mutations in human KCNE3 that disrupt its ability to inhibit Kv4.3 channels in human ventricle are thought to cause some cases of Brugada syndrome^74,126. Human KCNE3 mutations that cause gain-of-function of KCNQ1-KCNE3 channels are associated with atrial fibrillation, also suggested to arise from altered atrial myocyte channel activity¹²⁷. Human KCNE3 sequence variants associated with human disease are summarized in Table 3.

Table 3. KCNE3 gene variants associated or putatively associated with human disease.

AF, Atrial Fibrillation; BrS, Brugada Syndrome; fs, frame shift; LQTS, Long QT syndrome; n.r., not reported; Sources include cited papers and ¹⁶². KCNE3 Gene accession number: Q9Y6H6.

Mutation	Disease	In vitro cellular effects
T4A	LQTS? BrS	No effect on IQ1-E3 163 Ito gain of function126
V17M	AF	KV4.3 and Kv11.1 (hERG) gain of function 127
R53H	AF	IQ1-E3 gain of function164
F66F (198T/C synonymous switch)	Ménière’s disease (Japanese cohort; not recapitulated in study of Caucasian population) 122,123	n.r.
R99H	BrS LQTS?	KV4.3, Ito gain of function74 IQ1-E3 Loss of function163
R83H	Periodic paralysis (but also detected in asymptomatic individuals in a later study)65,120,121	Kv3.4 loss of function and increased sensitivity to proton block65

KCNE3 transcript was also detected in horse heart⁴³. However, Kcne3 was not detected in either the atria or ventricles of adult mice, by two different labs using different Kcne3^−/− mouse strains as negative controls^118,128, and Kcne3 deletion did not alter Kv currents in isolated ventricular myocytes¹²⁸. Yet, Kcne3 deletion predisposed to ventricular arrhythmogenesis in mice, lengthening the QT interval in older adults and predisposing to sustained ventricular tachycardias following ischemia/reperfusion¹²⁸. The underlying mechanism is unprecedented, and involves hyperaldosteronism in Kcne3^−/− mice associated with autoimmune attack on the adrenal glands. Kcne3 is not expressed in mouse adrenals, but its deletion causes cytokines to be released that attract activated lymphocytes specifically to the adrenal glands. Administration of spironolactone, an aldosterone receptor antagonist, to Kcne3^−/− mice ameliorated their QT prolongation and predisposition to ventricular tachycardia¹²⁸. Two questions arise: first, what causes the adrenals in Kcne3^−/− mice to release signals to attract activated B cells? Second, can this mechanism also be a component of human KCNE3-linked arrhythmogenesis? Resolution of these condundra will require further investigation.

Conclusions

In many ways, KCNE1 and KCNE3 represent the yin and yang of Kv channel regulation because of the often opposite yet complementary outcomes of their co-assembly with Kv α subunits. As more and more functional data emerge from studies of the KCNE family in expression systems and from mouse and human genetics studies, it becomes apparent that the KCNEs constitute a molecular toolkit that diversifies Kv channel functionality but also facilitates dynamic alteration of channel properties in response to other signaling elements – including membrane lipids, hormones, kinases and even larger membrane proteins such as solute transporters¹²⁹. Their human disease associations will likely not be as common as those of the Kv α subunits for a variety of reasons – although in LQTS, base-for-base they match up rather well to α subunit genes in terms of population-wide contribution to pathogenicity¹³⁰. Nevertheless, the phenotypes of Kcne null mice are fascinating and can inform us of their physiological roles and the consequences of their ablation (typically an altogether different lesion than a point mutation). Because KCNEs can alter Kv channel responses to native signals and also to pharmacological agents (in the context of both therapeutic targeting and side effects)^{11,16,130–133}, it behooves us to consider their presence in Kv and other channel complexes in different tissues, cell types and life stages, as they may yet provide a crucial ally in tuning drug specificity and efficacy.

Highlights.

• KCNE1 and KCNE3 regulate K⁺ channel gating, voltage dependence and trafficking

• KCNE1 and 3 often have opposite effects on channel function, especially so for KCNQ1

• KCNE1-KCNQ1 channels are best known for their role in cardiomyocyte repolarization

• KCNE3-KCNQ1 channels are important in various secretory epithelia

• Human KCNE1 and KCNE3 sequence variants cause various cardiac arrhythmias

Abbreviations

AV block

atrioventricular block

BLAST

basic local alignment search tool

cAMP

cyclic adenosine monophosphate

Cav channel

voltage-gated calcium channel

CHO

Chinese Hamster ovary

di

drug-induced

ERRγ

estrogen-related receptor γ

EST

expressed sequence tag

fs

frameshift

HCN

hyperpolarization-activated, cyclic nucleotide-gated

HEK

human embryonic kidney

I_Ks/IsK

slowly activating K⁺ current

JLNS

Jervell and Lange-Nielsen syndrome

KChIP

K⁺ channel interacting protein

Kv channel

voltage-gated potassium channel

LQTS

Long QT syndrome

MiRP

MinK-related peptide

MTS

methanethiosulfonate

Nav channel

voltage-gated sodium channel

PIP₂

phosphatidylinositol 4,5-bisphosphate

PKA

protein kinase A

PKC

protein kinase C

QTc

QT interval corrected for heart rate

SGN

spiral ganglion neuron

SNP

single nucleotide polymorphism

SQTS

Short QT syndrome

TdP

torsade de pointes

TEVC

two-electrode voltage clamp

TM

transmembrane