Single channel kinetic analysis of the cAMP effect on IKs mutants, S209F and S27D/S92D

ABSTRACT The IKs current is important in the heart’s response to sympathetic stimulation. β-adrenergic stimulation increases the amount of IKs and creates a repolarization reserve that shortens the cardiac action potential duration. We have recently shown that 8-CPT-cAMP, a membrane-permeable cAMP analog, changes the channel kinetics and causes it to open more quickly and more often, as well as to higher subconductance levels, which produces an increase in the IKs current. The mechanism proposed to underlie these kinetic changes is increased activation of the voltage sensors. The present study extends our previous work and shows detailed subconductance analysis of the effects of 8-CPT-cAMP on an enhanced gating mutant (S209F) and on a double pseudo-phosphorylated IKs channel (S27D/S92D). 8-CPT-cAMP still produced kinetic changes in S209F + KCNE1, further enhancing gating, while S27D/S92D + KCNE1 showed no significant response to 8-CPT-cAMP, suggesting that these last two mutations fully recapitulate the effect of channel phosphorylation by cAMP.


Introduction
The I Ks potassium current is involved in terminating the plateau phase of the cardiac action potential and becomes more important as sympathetic activity increases the heart rate [1]. Under sympathetic stimulation, the I Ks current is enhanced in a cAMP-dependent fashion, leading to faster action potential repolarization, which shortens the action potential and allows adequate time for ventricular refilling [1][2][3][4]. Mutations in the channel can give rise to long and short QT syndrome as well as familial atrial fibrillation [5][6][7].
A homotetramer of peptides encoded by the KCNQ1 gene (Q1) form the channel and are modulated by KCNE1 (E1) accessory subunits to produce the I Ks current. When E1 subunits are a part of the channel complex, the current is converted from one that activates rapidly and inactivates, to one with very slow activation, deactivation kinetics and no inactivation [8,9]. The number of these accessory subunits in the channel complex can vary between 1 to 4 units per channel [10], which allows the channel to be modulated and perhaps regulated by the number of E1s in the complex. Two residues in the N-terminus are phosphorylated by cAMP, S27 and S92 [11][12][13] and under sympathetic stimulation, the N-terminus of Q1 becomes phosphorylated by protein kinase A (PKA), which is part of a macromolecular complex bound to the C-terminal domain of Q1 [14].
Exogenous 8-CPT-cAMP decreases the voltage threshold for activation and increases the magnitude of I Ks , but only when the KCNE1 subunit and yotiao are present [1,13,[15][16][17][18]. Single channel recording has revealed the detailed changes in channel kinetics in the presence of 8-CPT-cAMP. We showed that the increase in current observed upon 8-CPT-cAMP addition is caused by channels opening more quickly, more often and to higher open sublevels, and suggested that these effects were caused by increased activation of the voltage sensor domains (VSDs) [15]. Mutant Q1 channels with enhanced or fully activated VS were used to characterize this effect.
In this addendum, we have further analysed the single channel kinetics of two Q1 mutants used in the original paper. One is S209F, a high open probability (Po) mutant with enhanced gating, where 8-CPT-cAMP has some effect on the subconductance occupancy. The other is a double phosphomimetic mutant (S27D/S92D), which appeared to show no response to 8-CPT-cAMP, but that has kinetic properties more similar to wild-type than the single phosphomimetic mutant, S27D.

Results and discussion
Kinetic analysis of 8-CPT-cAMP on an enhanced gating mutant, S209F S209F is a kinetic gain-of-function mutation, which can be described as having a preferentially activated VSD stabilizing the open state. When complexed with E1, it has a Po between 0.6 and 0.7 [15,19], which is about 4x that of wild-type Q1 + E1 (0.15-0.2) [10,19,20]. In our previous paper [15], at the single channel level, there was no change in the Po (control 0.71 and 0.74 in 8-CPT-cAMP at + 60 mV [15]), but there was a reduction in the number of closed events and some changes in subconductance levels from the all-point histogram distribution in the presence of 8-CPT-cAMP.
To further investigate S209F changes in subconductance in response to 8-CPT-cAMP, analysis of 20 S209F + E1 sweeps before and after 8-CPT-cAMP was carried out, and shows changes in the substate occupancy levels in the presence of 8-CPT-cAMP ( Figure 1). The closed dwell time percentage in control is~44%, which is considerably less than EQ even in the presence of 8-CPT-cAMP (Control 88% and 8-CPT-cAMP 79%, Thompson et al [15]. This closed dwell time drops to~21% in the presence of 8-CPT-cAMP. The mean closed dwell time duration are also significantly shorter in the presence of 8-CPT-cAMP (p-value: < 0.0001). The two highest sublevels 0.5 and 0.75 pA both increased in total dwell time by~18 and 9% respectively (Figure 1(c)), showing a shift in occupancy to higher sublevels. There are no significant changes in the mean dwell times at any of the open levels (p-value: > 0.999) (Figure 1(c)).
The S209 residue is buried in the hydrophobic core in the open state. The mutation converts a polar uncharged residue into a bulky hydrophobic one, which is thought to stabilize the open state and destabilize the closed state, resulting in gain-of-function activity [21]. This mutant can presumably gate normally as the voltage sensor is not "locked" unlike the E160R/R237E (E1R/R4E) mutant, which is locked in an activated state and appears to be unaffected by 8-CPT-cAMP [14]. While S209F has been described as "fully" activated, 8-CPT-cAMP does appear to further stabilize the open state by increasing events at higher subconducting states ( Figure 1). A reduction to~20% in closed time in the presence of 8-CPT-cAMP may also suggest a destabilization of the closed state (Figure 1(c)).
Data in Figure 2(c) show the probability distribution plot of burst durations before and after 8-CPT-cAMP. There is a small decrease in the number of short and intermediate bursts (Figure 2(c)). The open dwell times in control could be fit with three time constants (τ 1 : 1.70 ± 0.09, τ 2 : 13.87 ± 2.41, τ 3 : 44.38 ± 12.75 ms), however the second time constant was absent in the presence of 8-CPT-cAMP, as only two time constants could be fit (τ 1 : 2.62 ± 0.29 and τ 2 : 41.35 ± 3.36 ms). 8-CPT-cAMP may be having a small effect on the pore as well as the VSD. However, the fits for these 8-CPT-cAMP data may reflect the presence of fewer events. Comparing the EQ burst probability distribution plots [15] to the S209F + E1 plots, S209F has a wider range of burst durations, whereas EQ has the majority of its events taking place at the shorter burst durations. S209F + E1 is still able to respond to 8-CPT-cAMP at the single channel level, with a reduction in the total closed dwell time and an increase in the time spent at the higher subconducting levels. This suggests that even though this channel has enhanced activation, it can be still further increased by cAMP.
Single channel analysis of Q1 doublephosphomimetic mutant S27D/S92D The S27 residue in the N-terminus of Q1 is an important phosphorylation site during adrenergic stimulation [13], which can be pseudo-phosphorylated by mutating the serine to aspartic acid, S27D [22]. This mutation produces currents of greater magnitude and with a more hyperpolarized V 1/2 [22], but can be further phosphorylated, which suggests that there are other potential significant phosphorylation residues [15]. S92 has been shown by others to be phosphorylated [11], so we investigated the importance of this residue by using a double phosphomimetic mutant (S27D/S92D) and tested its response to 8-CPT-cAMP.
At the whole-cell level, there was no change in current size or V 1/2 of activation in response to 8-CPT-cAMP. At the single channel level, there was also no change in first latency or any significant change in the all-points histograms (Figure 3(a) & [15]). We carried out further investigation by analyzing subconductance occupancy (Figure 3). The idealized histograms in Figure 3 With respect to the closed and open dwell times (Figure 4), two resolvable closed states were found for both control and 8-CPT-cAMP events. The faster time constant in control and cAMP was similar, 1.56 ± 0.06 and 1.24 ± 0.14 ms respectively (Figure 4(a)). The second time constant was slower in control 7.18 ± 1.03 compared to 3.44 ± 0.89 ms in 8- level, and the number of events at each threshold before and after 8-CPT-cAMP. These data were filtered at 0.5 kHz, and the bin width used was 0.01 pA. Only events longer than 1.5 ms were included. One-way ANOVA was used to compare control and 8-CPT-cAMP mean dwell times of each sublevel. The mean dwell times between control and 8-CPT-cAMP were significantly different (p-value < 0.0001). The p-values for all other sublevels were > 0.9998.
CPT-cAMP (Figure 4(a)). However, this could be due to a smaller number of events in the control fit. When the closed dwell times are plotted as a probability distribution (Figure 4 (Figure 4(c)).
In Xenopus oocytes, the Q1 S27D/S92D mutant appears to have a more hyperpolarized V 1/2 compared to wild-type [23], which is what one might expect of a phosphomimetic mutant. However, in mammalian cells this construct produced a V 1/2 that was more depolarized than wild-type (Q1 S27D/S92D + E1 V 1/2 : 47.5 ± 8.6 mV and Q1 + E1 V 1/2 : 25.1 ± 2.5 mV [15]). This was surprising, and interestingly from the single channel analysis (Figure 4), we can see that this mutant does not quite behave like wild-type either. There is no visible main active opening amplitude peak of~0.45 pA that is usually observed in wildtype channels (Figure 3(a)), although the first latency of the channel was similar to that of EQ (1.62 ± 0.08 s and 1.61 ± 0.13 s respectively) [15]. While this double pseudo-phosphorylation mutant does not appear to be affected kinetically by 8-CPT-cAMP, this mutant still showed altered single channel kinetics compared to wild-type.

Molecular biology
Constructs used were as previously described in Thompson et al. [15].

Cell culture and transfections
For single channel recording, ltk − mouse fibroblast (LM) cells were used. They were grown in Minimum Eagle Medium (Thermo Fisher Scientific) with 10% fetal bovine serum, 100 U/ ml penicillin and 100 µg/ml streptomycin level, and the number of events at each threshold before and after 8-CPT-cAMP. These data were filtered at 0.5 kHz, and the bin width used was 0.01 pA. Only events longer than 1.5 ms were included. A one-way ANOVA was used to compare control and 8-CPT-cAMP mean dwell times of each sublevel. None of the 8-CPT-cAMP mean dwell times were significantly different from control (p-value between closed sublevels was 0.1355, all other sublevels were > 0.9999).

Patch-clamp electrophysiology
Single-channel recording methodology was performed as previously described in Thompson et al. and Werry et al. [15,19]. Data were obtained and analyzed using Axopatch hardware and pCLAMP 10.5 software (Molecular Devices).

Solutions
The solutions used for both single-channel recordings were as described in Thompson et al [15].
were included in the analysis. One-way ANOVAs using Sidak's multiple comparisons test were performed using Prism 7 (GraphPad Software).

Statistics
Results shown here are mean values ± SE (Prism 7, GraphPad Software).