Disease-associated mutations in KCNE potassium channel subunits (MiRPs) reveal promiscuous disruption of multiple currents and conservation of mechanism

KCNE genes encode single transmem-brane-domain subunits, the MinK-related peptides (MiRPs), which assemble with pore-forming (cid:1) subunits to establish the attributes of potassium channels in vivo. To investigate whether MinK, MiRP1, and MiRP2 operate similarly with their known native (cid:1) subunit partners (KCNQ1, HERG, and Kv3.4, respectively) two conserved residues associated with human disease and inﬂuential in channel function were evaluated. As MiRPs assemble with a variety of (cid:1) subunits in experimental cells and may do so in vivo, each peptide was also assessed with the other two (cid:1) subunits. Inherited mutation of aspartate to asparagine (D 3 N) to yield D76N-MinK is linked to cardiac arrhythmia and deafness; the analogs D82N-MiRP1 and D90N-MiRP2 were studied. Mutation of arginine to histidine (R 3 H) to yield R83H-MiRP2 is associated with periodic paralysis; the analogs K69H-MinK and K75H-MiRP1 were also studied. Macroscopic and single-channel currents showed that D 3 N mutations suppressed a subset of functions whereas R/K 3 H changes altered the activity of MinK, MiRP1, and MiRP2 with all three (cid:1) subunits. The ﬁndings indicate that the KCNE peptides interact similarly with different (cid:1) subunits and suggest a hypothesis: that clinical manifestations of inherited KCNE point mutations result from disruption of multiple native currents via promiscuous interactions.—

and pharmacology (2). Subunits encoded by KCNE genes are single transmembrane-domain peptides that assemble with ␣ subunits to influence all these functional parameters (3). Thus, the KCNE1 product MinK (4) partners with the ␣ subunit KCNQ1 to form the slowly activating cardiac current I Ks found in heart and ear (5,6). Similarly, the KCNE2 product MinK-related peptide 1 (MiRP1) assembles with the ␣ subunit HERG to generate a current resembling cardiac I Kr (7) and the KCNE3 product MiRP2 combines with the Kv3.4 ␣ subunit to form a subthreshold, A-type channel in skeletal muscle (8). Inherited mutations in MinK are associated with cardiac arrhythmia and deafness (9 -12), those in MiRP1 with inherited and drug-induced cardiac arrhythmia (7,13), and those in MiRP2 have been linked to periodic paralysis (8).
The idea that KCNE peptides can alter the function of more than one ␣ subunit type suggests they might operate in different complexes in similar fashion, presumably via common subunit-subunit interactions. To explore this notion, we studied two residues essential to MinK and MiRP2 function and strongly conserved in the five known MiRPs: D76N-MinK was recognized by its association with inherited arrhythmia and deafness and alters function of channels formed with KCNQ1; an equivalent aspartate (D) is present in all but MiRP3 (2,9,12,22). R83H-MiRP2 was identified by its sway over channels formed with Kv3.4 and association with periodic paralysis; an homologous arginine (R) or lysine (K) is found in all five MiRPs (2,3,8). Here, despite changes at these conserved sites in MinK, MiRP1, and MiRP2, each peptide retained the capacity to form functional complexes with its ␣ subunit partners (KCNQ1, HERG, and Kv3.4); conversely, D 3 N changes eliminated MiRP-mediated current up-regulation (but not down-regulation) and R/K 3 H mutations suppressed effects on current activation, deactivation, and magnitude, suggesting this residue may have a key role in activity. The findings demonstrate that many MiRP/␣ subunit operations do proceed by shared mechanisms. The results suggest a hypothesis for how a single KCNE point mutation might cause disease-through effects on more than one ␣ subunit channel type in vivo.

Heterologous expression
Mutations were produced in wild-type human KCNE2 and KCNE3 in pGA1 (7) or wild-type human KCNE1 (Ser38 variant) in pRAT (23) by pfu-based mutagenesis (Quick-Change Kit; Stratagene, La Jolla, CA), followed by insertion of mutant gene fragments into translationally silent restriction sites and confirmed by DNA sequencing. For studies in oocytes, cRNA was produced as before (7). Chinese hamster ovary (CHO) cells were treated with Superfect (Qiagen, Chatsworth, CA) and plasmids carrying the genes for MIRP2, Kv3.4, and green fluorescent protein were expressed in pCI neo-based vectors (Promega, Madison, WI), as before (8).
Oocytes HERG, KCNQ1, or Kv3.4 channels with and without MinK, MiRP1, or MiRP2 were assessed using Xenopus laevis oocytes and two-electrode voltage clamp (TEVC), as before (7). Resting membrane potential (RMP) of oocytes was assessed at steady state in current clamp. All experiments were performed at room temperature. Oocytes were used for most subunit partnerships. MiRP2/Kv3.4 channels were usually studied in CHO cells, as regulatory influences maintain stable current levels over the course of measurements (not shown). That MiRP2/Kv3.4 complexes do assemble in oocytes is shown by altered affinity of BDS-II here and earlier (8). Such cell-specific effects of KCNE peptides on current density and pharmacology are not without precedent; see refs 14, 24 for differential effects of D76N-minK on HERG and (7) for altered E-4031 blockade, respectively. For BDS-II blockade, CHO cell recordings were performed using outside-out patches, oocyte recordings using whole-cell TEVC; BDS-II was applied via a bath in both cases.

Protocols
Holding voltage was Ϫ80 mV unless stated otherwise. 1: Peak patch current; 3 s pulse from Ϫ100 to 90 mV in 10 mV steps with a 10 s interpulse interval. 2: Single voltage/drug inhibition; repetitive 3 s pulses to a single voltage as indicated with a 5 s interpulse interval. 3: Steady state; holding at Ϫ40 mV for 5-20 min with no interpulse interval. 4: Peak whole-cell current; 3 s pulse from Ϫ120 to 60 mV in 20 mV steps, followed by a 1.5 s tail pulse to Ϫ30 mV, with a 10 s interpulse interval. 5: Peak whole-cell current (MinK); 5 s pulse from Ϫ80 to 60 mV in 20 mV steps with a 10 s interpulse interval. 6: Steady-state activation; 3 s prepulse from Ϫ80 to 60 mV in 20 mV steps, followed by a 3 s pulse to Ϫ30 mV, with a 10 s interpulse interval. 7: Peak tail current/deactivation kinetics; 3 s prepulse to 40 mV, followed by a 3 s pulse from Ϫ120 to 60 mV in 20 mV steps with a 10 s interpulse interval.

Ionic conditions
Pipette solution for CHO cells in on-cell mode was (in mM): 100 KCl, 0.7 MgCl 2 , 1 CaCl 2 , 10 HEPES pH 7.4. CHO cell bath solution was as in the pipette. Bath solution for oocytes was (in mM) 4 KCl, 96 NaCl, 0.7 MgCl 2 , 1 CaCl 2 , 10 HEPES pH 7.4; for RMP measurements, KCl was varied from 0.01 to 100 mM, with NaCl being substituted to keep the total monovalent ion concentration at 100 mM.

A conserved KCNE site influential in both MinK/KCNQ1 and MiRP2/Kv3.4 channels
MinK assembles with KCNQ1 to slow channel activation, alter the voltage dependence of gating, and increase unitary conductance, thereby producing the attributes of native I Ks channels (5,6). Aspartate (D) to asparagine (N) mutation to yield D76N-MinK is associated with inherited long QT syndrome, sudden death, and deafness (9 -11). Compared with wild-type MinK/ KCNQ1 channels, the mutant reduces current density due to a shift of voltage-dependent activation to more positive potentials and a fourfold reduction in unitary conductance (9,12). Sequence alignment of MinK, MiRP1, and MiRP2 indicates that D76-MinK is a conserved residue (Fig. 1); a homologous D is also present in MiRP4 but not MiRP3 (3). To assess whether this site was significant to MiRP2 function, D90N-MiRP2 was produced and studied with its skeletal muscle partner Kv3.4.
Whenever KCNE mutation yielded attributes like those for channels formed by ␣ subunits alone, we sought evidence to confirm that the altered peptides were assembled into the channel complex. Two observations with macroscopic currents indicated that channels did contain D90N-MiRP2 subunits. First, the slope of the conductance-voltage relationship with D90N-MiRP2 was similar to that for wild-type MiRP2 (13Ϯ2.0 and 14Ϯ1.4, respectively) rather than that for homomeric Kv3.4 channels (22.5Ϯ0.6) (Fig. 2C). Second, channels with D90N-MiRP2 were blocked by the peptide toxin BDS-II like those containing wild-type MiRP2 (K i ϭ633Ϯ110 and 688Ϯ166 nM atϩ30 mV, respectively, nϭ4 -5 cells), and significantly different from channels formed with Kv3.4 subunits alone, which showed ϳsixfold greater sensitivity to the toxin (99.5Ϯ32 nM) (Fig. 2D).
At the single-channel level, the effects of D76N-MinK on KCNQ1 and D90N-MiRP2 on Kv3.4 were also similar ( Fig. 3). 1) D90N-MiRP2 produced a shift in activation voltage such that openings were rare and short-lived at voltages negative to Ϫ20 mV (Fig. 3A).     Fig. 3E, lower). Although dwell-time statistics for D90N-MiRP2/Kv3.4 channels should be treated with caution due to infrequent openings at this voltage, incorporation of the mutant into the complex was readily appreciated because single Kv3.4 channels demonstrate short-lived dwell times best fit to just a single exponential, as reported ( ϭ 2.60Ϯ0.06 ms) (8).
The D90N-MiRP2 subunit also acted to decrease the influence of MiRP2/KCNQ1 channels on RMP. The RMP of oocytes expressing only wild-type MiRP2 was not significantly different from control cells whereas expression of MiRP2 and KCNQ1 produced a shift of Ϫ72 Ϯ 3 mV, D90N-MiRP2 and KCNQ1 a shift of Ϫ36 Ϯ 7 mV, and KCNQ1 alone a Ϫ22 Ϯ 2 mV change when evaluated with a low potassium bath solution (Fig. 4C).

R83H-MiRP2 affects KCNQ1 and Kv3.4 similarly
MiRP2 assembles with Kv3.4 to form a potassium channel active at subthreshold potentials that influences skeletal muscle cell RMP; the missense mutant R83H-MiRP2 is associated with an inherited periodic paralysis; compared with wild-type MiRP2, the mutant reduces currents and causes skeletal muscle cell depolarization (8). Three effects of the mutation on Kv3.4-containing channels are described here.
The effects of R83H-MiRP2 on KCNQ1 were similar to those of D90N-MiRP2 although less potent (Fig. 6). Like channels with wild-type MiRP2, R83H-MiRP2/ KCNQ1 exhibited a largely linear current-voltage relationship; however, the mutation restored some time and voltage dependence to channel activation and decreased macroscopic currents (Fig. 6A, B). Flux was decreased ϳtwofold from 7.3 Ϯ 0.4 A for channels with wild-type MiRP2 to 4.3 Ϯ 0.3 A for channels with R83H-MiRP2 (Fig. 6B), similar to what is seen with homomeric KCNQ1 channels (Fig. 4B). As a result, the RMP of R83H-MiRP2/KCNQ1-expressing cells was shifted by ϩ15 Ϯ 7 mV in comparison to those with wild-type MiRP2/KCNQ1 channels (Fig. 6B, inset). Confirmation that R83H-MiRP2 subunits remained competent to assemble with KCNQ1 was afforded by clotrimazole, which blocked homomeric KCNQ1 channels less effectively (K i ϭ419Ϯ22 M) than either those with wild-type or R83H-MiRP2 (K i ϭ217Ϯ20 and 177Ϯ18 M, respectively) (Fig. 6C).

R/K 3 H suppresses all KCNE effects on HERG whereas D 3 N impedes only up-regulation
MiRP1 assembles with HERG to pass currents with the attributes of cardiac I Kr channels, and inherited vari-ants of MiRP1 are associated with congenital long QT syndrome and drug-induced torsades de pointes (7,13). MiRP1 mutations associated with disease reduce currents passed through channels formed with HERG by decreasing unitary conductance, shifting the V 1/2 of activation, speeding deactivation, and/or increasing sensitivity to drug blockade. HERG was studied here with K75H-MiRP1 and D82N-MiRP1, changes analogous to those associated with disease when carried in MiRP2 and MinK; then R83H-MiRP2 and D90N-MiRP2 were evaluated.

Common effects of conserved sites in different MiRP/␣ subunit complexes
Two conserved residues have been studied to assess how KCNE peptides (MiRPs) alter ␣ subunit function and to determine whether these sites influence different ␣ subunits similarly. The sites were chosen because their inheritance in mutant form has been associated with disease and altered channel function. Mutation of these residues in MinK, MiRP1, or MiRP2 was found to have similar effects in channels formed with KCNQ1 or Kv3.4. Each wild-type MiRP acted with these ␣ subunits to increase potassium current, and both D3 N changes (MinK-76, MiRP1-82, and MiRP2-90) and R/K3 H changes (MinK-69, MiRP1-75, and MiRP2-83) disrupted this capacity to increase flux. Like arrhythmiaassociated D76N-MinK with KCNQ1 (12), D90N-MiRP2 shifted activation of Kv3.4 channels to more depolarized potentials, diminished steady-state open probability, lowered unitary conductance, and altered the voltage dependence KCNQ1 channels to diminish current density. Like periodic paralysis-associated R83H-MiRP2 with Kv3.4 (8), R83H-MiRP2 and K69H-MinK decreased current through KCNQ1 channels due to a shift in activation to more depolarized potentials.
HERG ␣ subunits are stimulated by assembly with MinK and suppressed by MiRP1 and MiRP2 (14,20) and the two conserved residues exhibited differential effects on these regulatory events. Whereas stimulation of HERG was impeded by D77N mutation in rat minK (14), D82N-MiRP1 and D90N-MiRP2 suppressed activity like wild-type subunits. Conversely, K75H-MiRP1 and R83H-MiRP2 suppressed HERG less effectively than did wild-type.
Thus, R/K3 H mutation disrupted the ability of all three KCNE peptides to alter ␣ subunit function whether the wild-type effect was to increase or suppress current. This residue appears to be important for normal MiRP function. Whereas D3 N mutation disabled the ability of the MinK to increase potassium flux, it did not alter MiRP1 or MiRP2 suppression of HERG, suggesting either a more specific role for this residue or that the unique attributes of HERG yield this behavior. These effects of KCNE peptide mutations indicate their operation by common mechanisms and shared structures.
Sites of contact between MiRPs and ␣ subunits remain to be identified. Studies using mutagenesis, modification, pharmacology, electrophysiology, and antibody binding have revealed that KCNE peptides are type I transmembrane proteins that appear to cross the membrane in proximity to the channel pore (2, 7, 8, 22, 26). The effects of disease-associated mutations support a general correlation of function and domain: mutations in the extracellular region of MiRP1 alter voltagedependent activation and drug blockade (T8A and Q9E) (7,13); mutations in the transmembrane and membrane-following segments influence gating kinetics and ion conduction (7)(8)(9)12). Indeed, transplantation of residues 57-59 from the transmembrane segment of MinK to the equivalent sites in MiRP2 was sufficient to convert MiRP2/KCNQ1 channels from leak-type conductances to slowly activating MinK/ KCNQ1-like channels (19). The two sites studied here reside in the membrane-following region, and their potent influence on gating and pore function suggests an interaction with ␣ subunit S4, S5, and/or S6 segments (27,28).

Promiscuous interactions: a role in disease?
We hypothesize here that interaction of MiRPs with a variety of ␣ subunits and the capacity of MiRP mutations to alter function in the different complexes contribute to the varied clinical manifestations of MiRP-associated disease. Although the idea is appealing in its simplicity, proof for the hypothesis remains scant. A multiplicity of MIRP/␣ subunit interactions has been suggested by studies of cloned subunits in experimental cells, but in only a few cases has a direct comparison of cloned and native currents, transcript and/or protein expression, and/or linkage to human disease provided strong support for native assembly; these include MinK with KCNQ1 in heart and inner ear, MiRP1 with HERG in the heart, and MiRP2 with Kv3.4 in skeletal muscle (5-8, 29, 30).
Indeed, there is no compelling need to invoke promiscuous interactions in described KCNE-associated disorders. Cardiac arrhythmia and deafness in associa-tion with MinK mutations can be fully rationalized by diminished function of MinK/KCNQ1 channels in both the heart and ear (29,(31)(32)(33). MiRP1 mutations associated with a prolonged QT interval at baseline (Q9E, M54T, I57T, and A116V) act as one might expect to reduce currents passed through channels formed with HERG (by decreasing unitary conductance, shifting the voltage dependence of activation, and/or speeding of deactivation, thereby slowing cardiac repolarization) (7,13). These rare mutations create a substrate for a poor response to further current suppression by drugs known to block cardiac potassium channels (13). More specific effects are observed with one rare mutation (Q9E) and a common polymorphism (T8A) linked to antibiotic-induced arrhythmia, as these MiRP1 variants increase sensitivity of MiRP1/ HERG channels to blockade by the inciting drugs (7,13). Finally, R83H-MiRP2 is associated with periodic paralysis and alters both the attributes of a prominent skeletal muscle cell potassium current and the resting membrane potential of muscle cells via endogenous MiRP2/Kv3.4 channels (8).
Other observations remain controversial. We reported association of MiRP1 mutations with cardiac arrhythmias (inherited and drug-induced) and altered function of MiRP1/HERG channels (7,13); others have subsequently suggested that MiRP1 interaction with KCNQ1 (34), HCN1 (18), or Kv4.2 (17) might instead produce abnormal cardiac rhythms. So far we have been unable to demonstrate an effect of MiRP1 on KCNQ1 or HCN1 currents in either oocytes or CHO cells; evidence for these interactions may require specific experimental conditions. We do observe MiRP1 to alter Kv4.2 subunit function (16), but suspect that this influences the central nervous system in humans rather than the heart as Kv4.2 transcript is abundant in the brain but appears to be absent from human heart (35). Similarly, patients with periodic paralysis in association with MiRP2 mutation show abnormal skeletal muscle cell function reflective of altered MiRP2/Kv3.4 channel activity (8); others speculate that the primary role for MiRP2 is with KCNQ1 in the colon (based on transcript expression patterns) (20). Although notable effects of MiRP2 on KCNQ1 are reproduced here, individuals with MiRP2-associated periodic paralysis did not show abnormal gastrointestinal function (8) as might be expected if MiRP2/KCNQ1 complexes were central to fluid handling in the colon (20). Moreover, two probes used to demonstrate abundant transcript for MiRP2 in human skeletal muscle failed to visualize mRNA in the colon (8). Finally, we could not recapitulate MiRP2 suppression of KCNQ4 currents (20) when the two subunits were coexpressed in oocytes (not shown).
Why, then, hypothesize a role for promiscuous interaction? First, there is marked variation among recognized cases of KCNE-associated disease in clinical presentation, response to medical intervention, and longterm outcome, which may reflect such additional MinK, MiRP1, or MiRP2 interactions. Second, function of KCNE peptides with multiple ␣ subunit types might be difficult to discern if assembly takes place only in some cell types or under unique circumstances; indeed, cardiac potassium channel ␣ subunits show spatial and temporal variation in expression during development and in disease (36). Finally, it is the very capacity of KCNE peptides to operate with different ␣ subunits in experimental cells that impels consideration of the hypothesis that the interactions occur in nature. Proof requires discovery of a KCNE mutation that yields pathology due to its aberrant function in more than one channel type.