Human voltage-gated potassium (Kv) channels are generated by a gene family expressing 40 different α subunits, producing a wide range of current properties and diversity of conductances, gating kinetics and voltage dependence. Kv channels exist as tetramers that form a functional pore, and channel function can be diversified in some cases by formation of heteromeric channels bearing more than one type of α subunit, typically within the same Kv subfamily (Xu et al., 1995). Another mechanism for amplification of this diversity is splice variation (Vega-Saenz de Miera et al., 1992).

Kv4 subfamily α subunits (Kv4.1, Kv4.2, and Kv4.3), encoded by three genes (KCND1, 2, and 3, respectively), generate subthreshold-activating Kv currents that are transient because they exhibit fast inactivation, causing current decay over tens to hundreds of milliseconds (Serôdio et al., 1994, 1996). Kv4.3 is strongly expressed in human cardiomyocytes and neurons. In neurons, Kv4.3 contributes to I_SA (Serôdio et al., 1994), while in cardiac myocytes it generates I_to (transient outward current; Dixon et al., 1996; Johns et al., 1997), important for early myocyte repolarization to counteract depolarization caused by cation influx through voltage-gated sodium and calcium channels. Kv4.1, 2, and 3 exhibit close sequence alignment, except for a divergent additional exon in Kv4.3, beginning after residue 487, encoding 19 amino acids within the S6-proximal cytosolic C-terminus (residues 488–506) and unique to Kv4.3. This additional exon is commonly spliced such that two Kv4.3 splice variants exist in human tissues. The short version (Kv4.3S) lacks the 19 residues whereas the long form (Kv4.3L) possesses the additional 19 amino acids but is otherwise identical to Kv4.3S (Ohya et al., 1997; Kong et al., 1998).

Both forms of Kv4.3 are expressed in human and rat heart (Ohya et al., 1997; Kong et al., 1998) but their expression levels are altered differentially in human heart failure (HF), Kv4.3L expression rising ~33% while that of Kv4.3S falls ~75% (Radicke et al., 2006). HF has reached epidemic proportions, with a prevalence of >37.7 million people worldwide, and almost 6 million in the United States alone, half of whom die within 5 years of diagnosis. HF occurs when heart muscle becomes so weak that it cannot adequately pump blood into the aorta to supply the body with oxygenated blood, resulting in pooling of blood in the heart. HF risk increases with diabetes, obesity, hypertension and coronary artery disease, and lifestyle factors including a sodium- and cholesterol-rich diet, and lack of exercise (Roger, 2013; Mozaffarian et al., 2016; Ziaeian and Fonarow, 2016). HF is tightly associated with reduced I_to density and Kv4.3 expression in human and canine ventricular myocytes. A reduction in I_to alters cardiomyocyte action potential morphology and duration, particularly the early repolarization and plateau phases, at least partially by perturbing voltage-gated Ca²⁺ channel activity (Beuckelmann et al., 1993; Näbauer et al., 1993; Kääb et al., 1998; Zicha et al., 2004). Approximately half of individuals with HF die suddenly, with ventricular tachyarrhythmias being a likely cause in many cases (Cohn et al., 1986; Stevenson et al., 1993). A greater understanding of the processes underlying perturbation in HF of ionic currents, including I_to, known to be associated with ventricular tachyarrhythmias in other genetic arrhythmia syndromes, has the potential to lead to therapeutic advances in reducing morbidity and mortality in HF.

Thus, the prior finding that HF splice-dependently remodels Kv4.3 in human heart is potentially of high significance. However, homomeric Kv4.3L and Kv4.3S are functionally indistinguishable at baseline, in terms of macroscopic current density, gating kinetics, and voltage dependence. Subtle isoform-dependent differences arise because Kv4.3L contains a consensus phosphorylation site in its unique 19-residue stretch. Protein kinase C (PKC) phosphorylation of threonine 504 within this site differentially modulates Kv4.3L closed state inactivation, increasing its magnitude in Kv4.3L, while in Kv4.3S (lacking this site) PKC reduces closed state inactivation. Aside from this, other effects of PKC were found to be Kv4.3 isoform-independent, similarly inhibiting peak current and not altering open-state inactivation or recovery from open-state inactivation (Xie et al., 2009).

Kv currents are further diversified in vivo by formation of macromolecular complexes between Kv α subunits and other proteins that alter their functional characteristics. Many of these channel-regulating proteins appear to be bona fide channel components, forming stable, obligate, heteromeric channel complexes with α subunits in vivo, and are referred to as ancillary or β subunits (in the case of Kv channel modulators). Two major classes of Kv channel β subunits expressed in human heart are the KCNEs and the K⁺ channel interacting proteins (KChIPs; Abbott and Goldstein, 1998; An et al., 2000). The KCNEs (also termed MinK-related peptides, or MiRPs) are single-transmembrane-domain subunits that co-assemble with Kv α subunits to dictate fundamental properties including channel α subunit composition, forward trafficking, endocytosis, ion selectivity, conductance, gating kinetics, voltage dependence, and the effects of regulation by other proteins (Abbott, 2015, 2016a,b). KChIPs are cytosolic proteins that also modulate multiple aspects of Kv channel biology, including trafficking, current density, gating kinetics, and voltage dependence (An et al., 2000; Rhodes et al., 2004).

Kv4 channels are able to co-assemble with KCNE proteins and with KChIPs, almost certainly at the same time, forming heteromeric channels with varied gating properties and current magnitude (Radicke et al., 2006; Liu et al., 2008; Levy et al., 2010). It is thought that Kv4.3 is regulated in human heart by KChIP2 (and in particular the dominant cardiac isoform, KChIP2b) and one or more of the KCNEs (Kuo et al., 2001; Deschênes and Tomaselli, 2002; Radicke et al., 2006). Indeed, cardiac arrhythmias including Brugada syndrome have been associated with increased ventricular I_to arising from gain-of-function mutations in KCND3 (Giudicessi et al., 2011), but also in KCNE3 (Delpón et al., 2008) and KCNE5 (Ohno et al., 2011). Despite a plethora of studies on Kv4.3 regulation by KCNEs and KChIPs on the one hand, and of the function and expression of Kv4.3L vs. Kv4.3S on the other hand, a comparison of how the two Kv4.3 splice variants respond to regulation by β subunits has been lacking, and this response was assumed to be isoform-independent (Radicke et al., 2006). Here, it is demonstrated that KChIP2b and KCNEs unexpectedly introduce a number of striking differences in baseline function between human Kv4.3L and Kv4.3S. These data both identify a 19-residue C-terminal stretch important for Kv4.3 regulation by β subunits, and suggest the functional and pathophysiological significance of differential remodeling of Kv4.3 splice variants in HF.

Native fast-inactivating Kv currents are generated predominantly by channels containing Kv1.4, Kv3.3, Kv3.4, or Kv4.x α subunits, all of which exhibit inherent fast-inactivation apparatus [N-type inactivation domains, although the mechanistic nuances are different for Kv4.x α subunits (Bärghaan et al., 2008; Dougherty et al., 2008; Barghaan and Bähring, 2009)]. In the case of Kv1 and Kv3 subfamily channels, heteromers of α subunits containing both N-type (fast-inactivating) and delayed rectifier (slow, C-type inactivating) α subunits can generate currents with intermediate inactivation kinetics, proportional to the number of inactivation domains available to plug the pore after channel opening. Regulatory subunits, including protein kinases and KCNEs, may further modify Kv1 and Kv3 channel inactivation rates, and other factors including voltage dependence, surface expression, and even α subunit composition itself (Abbott et al., 2001, 2006; Grunnet et al., 2003; Kanda et al., 2011a,b). Kv4 channels are different in that there are no delayed rectifier α subunits in the Kv4 subfamily, so the channels they form are always fast-inactivating. Kv4 channels are, however, subject to modulation by several classes of regulatory proteins, thought to occur concurrently in vivo in human cardiac and neuronal Kv4-based macromolecular complexes. The major new finding in the present study is that beta subunits unlock previously unrecognized functional differences in Kv4.3 splice variants known to be expressed in human heart. This is of mechanistic importance, and is also potentially relevant to cardiovascular disease, as it predicts that previously recognized, HF-associated changes in I_to subunit expression could have a robust functional impact, reducing I_to density by as much as 50%.

The prior assumption with respect to Kv4.3L and Kv4.3S splice variants was that their function differed only with respect to their regulation by PKC, via threonine 504 within the C-terminal 19-residue stretch (Po et al., 2001; Xie et al., 2009). Radicke and colleagues analyzed the functional effects of KChIP2 and KCNE subunits on Kv4.3L and also discovered HF-associated remodeling of Kv4.3 splicing and β subunits, but did not examine functional effects on Kv4.3S (Radicke et al., 2006). While Kv4.3L transcript was increased by 33% in HF, Kv4.3S transcript was decreased by 75%, resulting in a net reduction in total Kv4.3 transcript. KCNE4 was the highest-expressed KCNE subunit (128 fg/ng in normal heart; 75 fg/ng in HF) followed by KCNE2 (6.5 fg/ng in normal heart; 5.3 ng in HF). KCNE1 transcript was increased from 1.7 to 4.7 fg/ng in HF (Radicke et al., 2006).

Placed into the context of the prior findings for HF-associated subunit remodeling, the combination of subunit expression changes observed in the present study would be predicted to have complex but predominantly inhibitory effects on I_to. KCNE4 (the long variant) was shown here to strongly inhibit Kv4.3L-KChIP2b current but have much weaker inhibitory effects on Kv4.3S-KChIP2b. Although KCNE4 expression dropped ~40% in human heart failure, it was still the predominant KCNE subunit. Therefore, if expressed in the same myocytes as Kv4.3 and KChIP2b, KCNE4 would be more inhibitory if the balance of Kv4.3 isoforms shifted heavily toward Kv4.3L as previously observed (Radicke et al., 2006). This effect would be even greater if normal hearts express heteromeric Kv4.3L-Kv4.3S-KChIP2b-KCNE4, as these generate even larger currents than Kv4.3S-KChIP2b-KCNE4, and would be predicted to be replaced by the lesser current-passing Kv4.3L-KChIP2b-KCNE4 (Figure 6). This inhibitory effect would be further exacerbated by reduced KChIP2 expression in failing hearts, as KChIP2b protects Kv4.3L from strong inhibition by KCNE4. Reduced KCNE5 in failing hearts would be predicted to increase I_to, but this would be at least partially mitigated by the relatively stronger inhibitory effects of KCNE5 on Kv4.3L-KChIP2b channels compared to Kv4.3S-KChIP2b (Figure 7). Finally, KCNE1 upregulation in failing heart would be predicted to increase I_Ks [formed by KCNQ1 and KCNE1 (Barhanin et al., 1996; Sanguinetti et al., 1996)] and I_Kr [formed by hERG and one or more KCNE subunits, including KCNE1—which augments hERG current (McDonald et al., 1997)]. However, in the case of Kv4.3, we found that KCNE1 only augments the activity of channels comprising Kv4.3S and KChIP2b (Figure 3), so this would again mitigate augmentation of I_to by increased KCNE1.

Radicke and colleagues did not compare the effects of KCNE subunits on Kv4.3L in the absence of KChIP2, because in CHO cells they did not observe Kv4.3L currents in the absence of KChIP2 (Radicke et al., 2006). Nor has anyone reported comparing effects of β subunits on Kv4.3S vs. Kv4.3L, at least partly because the conserved Kv4 N-terminal was the acknowledged binding site for KChIPs (Bähring et al., 2001; Wang, 2008), whereas the Kv4.3L and Kv4.3S isoforms differ only in their C-termini. However, the C-terminal of closely-related α subunit Kv4.2 also mediates functional effects of KChIPs (Callsen et al., 2005; Han et al., 2006). The current work recapitulates this for Kv4.3 and refines it to the extent that a biologically and pathophysiologically relevant 19-residue segment in the Kv4.3 membrane-proximal C-terminal is shown to markedly influence the functional consequences of Kv4.3 co-assembly with two classes of β subunit.

Only select consequences of KChIP2b regulation of Kv4.3 function were attenuated by the 19-residue segment (current augmentation and slowing of inactivation), and others were augmented (larger shift in steady-state inactivation voltage dependence) or unaffected (equally robust increase in rate of inactivation recovery; similar shift from double- to single-exponential current decay kinetics during inactivation). Thus, the 19 residue segment is not suggested to interfere with Kv4.3-KChIP2b complex formation, instead selectively modulating functional consequences of the interaction. The finding here that a synthetic “L” peptide mimicking the 19-residue segment can acutely alter the effects of KChIP2 on Kv4.3S, altering current magnitude and inactivation kinetics to levels intermediate between those of Kv4.3S and Kv4.3L (Figure 8), strongly supports the premise that the 19-residue segment is directly involved in modulating some of the functional effects of KChIP2 on Kv4.3L. There may also be a reduction of the current augmenting effects of KChIP2b upon Kv4.2, which also lacks the 19-residue stretch unique to Kv4.3L, although this did not reach statistical significance. Effects of the L peptide on the function of Kv4.3S (and on Kv4.2) in complexes with KChIP2b are most likely to be via direct interaction with KChIP2, because the L peptide had no effects on channels in the absence of KChIP2. However, it is technically also feasible that presence of the L peptide, or the corresponding segment in the Kv4.3L α subunit, alters conformation of another domain within Kv4.3 in such a way that the α subunit responds differently to KChIP2b. Because effects were observed within 30 min of peptide injection, an alternative hypothesis, that the 19-residue segment only exerts influence on Kv4.3-KChIP2 complexes early in biogenesis, or dictates whether they co-assemble in the first place, is highly unlikely given that the turnover time of Kv4.3-KChIP2 complexes is much longer [e.g., t_1/2 of decay in pulse chase studies of 7–9 h (Jiang et al., 2009)].

Results described above, together with the observation here that cRNA ratios approximating those previously observed in human HF (Radicke et al., 2006) dramatically reduce total Kv4.3 current (Figure 9), strongly suggests that the relative balance of Kv4.3L and Kv4.3S is crucial to dictating I_to density and kinetics when co-assembled with β subunits known to regulate Kv4.3 in human heart, and raises the idea that therapeutic targeting to specifically upregulate Kv4.3S would be more beneficial in terms of I_to density restoration in HF than blanket Kv4.3 upregulation.

It is important to also recognize limitations of the current study. Xenopus oocytes were used for channel expression, partly because their robust nature ensures a large quantity of recordings can be completed in a short time, facilitating direct comparison of many different subunit combinations within the same timeframe and same batch of oocytes. In addition, because Xenopus oocytes are injected individually with cRNA rather than transfected en masse with cDNA, one can be more assured that all the intended subunits are expressed in each, and in the intended quantities. However, no expression system is perfect, and Xenopus oocytes may lack some of the factors influencing I_to current characteristics in human heart, or endogenously express additional proteins not present in cardiac myocytes (although this criticism can also be leveled for immortal cell lines derived from mammals). Indeed, because of this, there are no claims herein that specific subunit combinations/splice variants better reflect native I_to than previous reports. Rather, a minimalist system has been utilized to uncover major, unanticipated Kv4.3 splice isoform-dependent functional differences at baseline. A comparison of different KChIP2 splice variants was not included here, and only the recently cloned long forms of KCNE3 and KCNE4 (Abbott, 2016c) were utilized, in favor of the short forms. In addition, other subunits that may also regulate Kv4.3 in human heart were not studied here. Thus, the experiments may fail to replicate other nuances that shape human cardiac I_to. Neither were effects of Brugada syndrome-linked mutations in Kv4.3 (KCND3; Giudicessi et al., 2011), KCNE3 (Delpón et al., 2008) and KCNE5 (Ohno et al., 2011) quantified here in the context of isoform-specific effects, but it will be interesting to investigate potential differences in the future. Finally, effects of protein kinases were considered outside the scope of this study, but a subject for future work. An interesting contrast between the Kv4.3 isoform-dependent effects observed here compared to those previously reported for PKC-derived effects is that for the latter, Kv4.3L characteristics dominated when the two isoforms were co-expressed and PKC activated (Po et al., 2001); in the present study, Kv4.3S characteristics presided (and were apparently further augmented) when co-expressed with Kv4.3L, KChIP2b and KCNE4 (Figure 6).

The author confirms being the sole contributor of this work and approved it for publication.

This work was supported by University of California, Irvine School of Medicine setup funds.