An Arrhythmic Mutation E7K Facilitates TRPM4 Channel Activation via Enhanced PIP2 Interaction

A Ca2+-activated monovalent cation-selective TRPM4 channel is abundantly expressed in the heart. Recently, a single gain-of-function mutation identified in the distal N-terminus of the human TRPM4 channel (Glu5 to Lys5; E7K) was found to be arrhythmogenic because of enhanced cell membrane expression. In this study, we conducted detailed analyses of this mutant channel from more functional aspects, in comparison with its wild type (WT). In an expression system, intracellular application of a short soluble PIP2 (diC8PIP2) restored the single-channel activities of both WT and E7K, which had quickly faded after membrane excision. The potency (Kd) of diC8PIP2 for this recovery was stronger in E7K than its WT (1.44 vs. 2.40 μM). FRET-based PIP2 measurements combined with the Danio rerio voltage-sensing phosphatase (DrVSP) and patch clamping revealed that lowering the endogenous PIP2 level by DrVSP activation reduced the TRPM4 channel activity. This effect was less prominent in E7K than its WT (apparent Kd values estimated from DrVSP-mediated PIP2 depletion: 0.97 and 1.06 μM, respectively), being associated with the differential PIP2-mediated modulation of voltage dependence. Moreover, intracellular perfusion of short N-terminal polypeptides containing either the ‘WT’ or ‘E7K’ sequences respectively attenuated the TRPM4 channel activation at whole-cell and single-channel levels, but in both configurations, the E7K polypeptide exerted greater inhibitory effects. These results collectively suggest that N-terminal interaction with endogenous PIP2 is essential for the TRPM4 channel to function, the extent of which may be abnormally strengthened by the E7K mutation through modulating voltage-dependent activation. The altered PIP2 interaction may account for the arrhythmogenic potential of this mutation.


Introduction
Despite steady advancements in prevention and treatment in the past few decades, cardiac arrhythmias remain among the leading causes of mortality in developed countries [1]. Thus, comprehensive understanding their pathogenesis continues to be an important goal with the highest priority. Recently, transient receptor potential (TRP) channels have drawn increasing attention because of their active involvement in cardiac pathophysiology including arrhythmias [2,3]. A notable example is the melastatin subfamily member TRPM4, which functions as a Ca 2+ -activated monovalent cation-selective cation channel [4]. It has been suggested that TRPM4 contributes to the prolongation of action potential (AP) and generation of early afterdepolarization (EAD) [5]. In clinical studies, the N-terminal mutation c.19G → A in the TRPM4 gene (i.e., p.7E → K) was found in a few pedigrees of

Solution
The standard external solution contained (in mM): 140 NaCl, 5 KCl, 1 CaCl 2 , 1.2 MgCl 2 , 10 HEPES, and 10 glucose (pH 7.4, adjusted with Tris base). The pipette solution for the whole-cell and inside-out (I/O) recordings contained (in mM): 120 Cs-aspartate, 20 CsCl, 2 MgCl 2 , 5 EGTA, 10 Hepes, 2 ATP, 0.1 GTP, and 10 glucose (adjusted to pH 7.2 with Tris base), to which an optimal amount of Ca 2+ was added to give a desired [Ca 2+ ] value. The [Ca 2+ ] value of the pipette solution was calculated by using a custom-written program based on Fabiato and Fabiato's algorithm written in Visual Basic with the enthalpic and ionic strength corrections of the association constants [17].

Electrophysiology
The whole-cell and I/O modes of the patch-clamp technique were employed for the membrane current recording. Electrodes with the input resistance of 4-6 MΩ (when filled with internal solution) were made from 1.5-mm borosilicate glass capillaries (Sutter Instrument), and connected to the headstage of a low-noise, high-impedance patch-clamp amplifier (EPC10, HEKA Elektronik, Ludwigshafen, Germany). The amplifier was controlled by the automated multichannel data acquisition software 'Patchmaster' (HEKA, Ludwigshafen, Germany). In total, >60% of series resistance was electronically compensated. For continuous monitoring, current and voltage signals were sampled via the PowerLab data acquisition system (AD Instruments, Sydney, Australia) and subjected to subsequent offline analyses. For kinetic analyses and illustrations, data exported in the matlab or text formats were processed by the commercial data analysis software Origin v.9.1 (LightStone, Tokyo, Japan), Clampfit v.10 (Axon Instruments, Foster City, CA, USA), Excel 2012, and KaleidaGraph v.4 (Hulinks, Tokyo, Japan). To activate DrVSP, depolarizing step pulses (from the holding potential of −60 mV to 120 mV) of varying durations were applied every 120 s. The ratio of the inward currents after (I post ) to before (I pre ) DrVSP activation was used to calculate the degree of DrVSP-mediated inhibition according to the equation: 1.0-I post /I pre .

FRET Measurement
The method used for FRET measurement was essentially the same as described previously [16]. In brief, fluorescence emissions from voltage-clamped cells were obtained via a high-sensitivity sCMOS camera (Andor Neo; Andor Technology) equipped to a microscope (IX71; Nikon, Tokyo, Japan). Excitation lights filtered at 430 ± 10 and 504 ± 12 nm, respectively, were alternately shone by a computer-controlled high-speed wavelength-switching light source (Lambda DG-4, Sutter Instrument Co., Novato, CA, USA). Epifluorescence was first prefiltered through a multiband dichroic mirror (449-483 and 530-569 nm) contained in the microscope, and then separated and filtered in a beam splitter (Dual-View2; Photometrics) at 464 ± 23 nm (for donor fluorescence) and 542 ± 27 nm (for acceptor fluorescence). The duration of the camera exposure was confined to 250 ms within the 300-ms period of illumination for each excitation wavelength. Captured images were digitized as 16-bit, 512 × 512 pixels using the imaging software 'MetaMorph v.7.7 . Averaged intensities from the whole-cell region were also calculated by this software to obtain the FRET. FRET and electrophysiological measurements were synchronized by a brief trigger output from the PowerLab (AD Instruments, Sydney, Australia) to the excitation light shutter.

Statistical Evaluation
All experimental data are expressed as the mean ± SEM. Statistical significance (p < 0.05) was evaluated by the Student's t-test or ANOVA with Tukey's or Dunnett's post hoc tests for single or multiple comparisons, respectively.

DiC 8 -PIP 2 More Potently Reactivates E7K Than WT-TRPM4 Channels
First, to confirm the previous finding that PI(4,5)P 2 is essential for TRPM4 channel activity [12,13], and to explore whether there is any difference in PIP 2 sensitivity between its WT and E7K mutant, we performed single-channel recordings of TRPM4 channels heterologously expressed in HEK293 cells. Both the WT-and E7K-TRPM4 channels underwent rapid desensitization/rundown upon membrane excision into a 300 µM Ca 2+ -containing bathing solution. The subsequent application of a water-soluble short form of PIP 2 , diC 8 -PI(4,5)P 2 (5 µM) quickly restored the channel activity in both channels. ( Figure 1A). To quantify these effects, we applied a broader concentration range of diC 8 -PIP 2 (0.2-20 µM) in an incremental manner ( Figure 1B). The extent of reactivation (see the Figure 1 legend) was well described by a Hill-type equation with apparent EC 50 values of 2.40 ± 0.23 µM (n = 5) and 1.44 ± 0.20 µM (n = 5) for the WT-and E7K-TRPM4 channels, respectively ( Figure 1C). These results indicate that the E7K mutant is more potently reactivated by diC 8 -PIP 2 than its WT.

Simultaneous Measurements of Endogenous PI(4,5)P 2 and TRPM4 Channel Activity
To explore the significance of the above finding in a more physiological context, we next investigated the functional correlation between the magnitude of the whole-cell TRPM4 current and endogenous PIP 2 level, on the basis of the same approach as adopted previously [16]. Specifically, depolarizing pulses were applied to activate DrVSP, and thereby dephosphorylate PIP 2 , in the cell membrane. To monitor the endogenous PIP 2 level, we employed CFPmse-PHd and YFPmse-PHd as a FRET donor/acceptor pair. Under unstimulated conditions, this pair is presumed to localize mostly in close vicinity at the inner cell membrane, thereby producing significant FRET, while after PIP 2 depletion, it is expected to dissociate from the membrane to decrease FRET. In addition, the whole-cell TRPM4 current was induced by the intracellular infusion of a moderate concentration of Ca 2+ (1 µM) via sharp patch electrodes in order to minimize the time-dependent desensitization/rundown, as performed previously [19]. Figure 2A demonstrates the typical recordings of the fluorescence intensities of CFPmse-PHd and YFPmse-PHd in response to a brief depolarizing pulse from −60 mV to +120 mV. Immediately after DrVSP activation by the depolarizing pulse (Figure 2Aa), the acceptor fluorescence by donor excitation (F 542(D ) decreased (Figure 2Ad), while the donor fluorescence by donor excitation (F 464(D ) increased (Figure 2Ab), and then both recovered slowly to the original levels. In contrast, the acceptor fluorescence by acceptor excitation (F 542(A) ) remained almost constant before and after DrVSP activation (Figure 2Ac). The reciprocal changes in F 542(D) and F 464(D) suggest that a substantial decrease occurred in the efficiency of FRET from CFPmse-PHd to YFPmse-PHd upon the depolarization-induced PIP 2 depletion. The slow recovery of the fluorescence intensities most likely reflects the replenishment of membrane PIP 2 by PIP-5-kinase [20]. After the correction of nonspecific fluorescence changes by the 3-cube method (see the Materials and Methods Section) [16,18], the value of the FRET ratio (FR), which would faithfully reflect the dynamic change of membrane PIP 2 level, showed a more prominent decrease and recovery over time (Figure 2Bc). The time course of this FR change exactly coincided with that of the TRPM4 current that was simultaneously recorded (Figure 2Bc). This temporal coincidence strongly supports that the TRPM4 channel activity is closely corelated with the endogenous PIP 2 level. and I respectively denote the amplitudes of TRPM4 currents reactivated by given ('X' μM) and maximal (20 μM) concentrations of diC8-PI(4,5)P2, and I[post-desens] is that of the basal TRPM4 current after desensitization to 300 μM Ca 2+ [13]. Averaged concentration-response curves for the TRPM4 channel reactivation by diC8-PI (4,5)

Simultaneous Measurements of Endogenous PI(4,5)P2 and TRPM4 Channel Activity
To explore the significance of the above finding in a more physiological context, we next investigated the functional correlation between the magnitude of the whole-cell TRPM4 current and endogenous PIP2 level, on the basis of the same approach as adopted previously [16]. Specifically, depolarizing pulses were applied to activate DrVSP, and thereby dephosphorylate PIP2, in the cell membrane. To monitor the endogenous PIP2 level, we employed CFPmse-PHd and YFPmse-PHd as a FRET donor/acceptor pair. Under unstimulated conditions, this pair is presumed to localize mostly in close vicinity at the inner cell membrane, thereby producing significant FRET, while after PIP2 depletion, it is expected to dissociate from the membrane to decrease FRET. In addition, the wholecell TRPM4 current was induced by the intracellular infusion of a moderate concentration of Ca 2+ (1 μM) via sharp patch electrodes in order to minimize the time-dependent desensitization/rundown, as performed previously [19]. Figure 2A demonstrates the typical recordings of the fluorescence intensities of CFPmse-PHd and YFPmse-PHd in response to a brief depolarizing pulse from −60 mV to +120 mV. Immediately after DrVSP activation by the depolarizing pulse (Figure 2Aa  and I respectively denote the amplitudes of TRPM4 currents reactivated by given ('X' µM) and maximal (20 µM) concentrations of diC 8 -PI(4,5)P 2 , and I [post-desens] is that of the basal TRPM4 current after desensitization to 300 µM Ca 2+ [13]. Averaged concentration-response curves for the TRPM4 channel reactivation by diC 8 -PI(4,5)P 2 are fitted by the Hill-type equation: 1/(1 + (EC 50 /[diC 8 PIP 2 ]) n . It gives EC 50 values of 2.40 ± 0.23 µM and 1.44 ± 0.20 µM; Hill coefficient (n) values of 1.5 and 1.2 for WT and E7K, respectively. * p < 0.05 with ANOVA followed by Tukey's post hoc tests (n = 5).

Differential PIP 2 Sensitivities of WT and E7K-Mutant TRPM4 Channels
In the next step, we further questioned whether WT-and E7K-TRPM4 channels have differential sensitivities to the endogenous PIP 2 level. Depolarizing pulses of incremental durations (300-2000 ms) were applied to activate DrVSP and deplete membrane PIP 2 in a graded fashion [21]. As demonstrated in Figure 3A, prolonging the pulse duration progressively decreased both the FR value and the whole-cell TRPM4 current amplitude (I m ). Notably, even though there was no difference in the maximum inhibition of current amplitude, the degree of TRPM4 current inhibition by a given duration of depolarization (or value of FR) was smaller in E7K-mutant channels than in the WT-channels (open vs. filled circles in Figure 3B). After converting the FR value into the endogenous PIP 2 level (see the Figure 3 legend), the relationship between the TRPM4 current inhibition and endogenous PIP 2 level demonstrated a higher PIP 2 sensitivity of E7K-mutant channels than the WT-TRPM4 channels ( Figure 3C), especially near the range of K d values. Moreover, the recovery from the inhibition was significantly faster in E7K-than WT-TRPM4 channels  Figure 3Da); the time constants of recovery were 13.71 ± 4.24 s and 17.56 ± 5.29 s for E7K and WT, respectively. These observations raise the idea that the E7K mutation may render the TRPM4 channel more tightly bound to endogenous PIP 2 , thereby stabilizing its activity.
gest that a substantial decrease occurred in the efficiency of FRET from CFPmse-PHd to YFPmse-PHd upon the depolarization-induced PIP2 depletion. The slow recovery of the fluorescence intensities most likely reflects the replenishment of membrane PIP2 by PIP-5-kinase [20]. After the correction of nonspecific fluorescence changes by the 3-cube method (see the Materials and Methods Section) [16,18], the value of the FRET ratio (FR), which would faithfully reflect the dynamic change of membrane PIP2 level, showed a more prominent decrease and recovery over time (Figure 2Bc). The time course of this FR change exactly coincided with that of the TRPM4 current that was simultaneously recorded ( Figure 2Bc). This temporal coincidence strongly supports that the TRPM4 channel activity is closely corelated with the endogenous PIP2 level.

Differential PIP2 Sensitivities of WT and E7K-Mutant TRPM4 Channels
In the next step, we further questioned whether WT-and E7K-TRPM4 channels have differential sensitivities to the endogenous PIP2 level. Depolarizing pulses of incremental durations (300-2000 ms) were applied to activate DrVSP and deplete membrane PIP2 in a graded fashion [21]. As demonstrated in Figure 3A, prolonging the pulse duration progressively decreased both the FR value and the whole-cell TRPM4 current amplitude (Im). Notably, even though there was no difference in the maximum inhibition of current amplitude, the degree of TRPM4 current inhibition by a given duration of depolarization (or value of FR) was smaller in E7K-mutant channels than in the WT-channels (open vs. filled circles in Figure 3B). After converting the FR value into the endogenous PIP2 level (see the

PIP 2 Depletion Only Modestly Affects the Voltage-Dependent Activation of E7K Mutant Because of Its Higher PIP 2 Affinity
It is well known that TRPM4 channels are positively regulated by the membrane potential. Although the mechanism of actions remains unclear, application of diC 8 -PIP 2 was found to cause a remarkable leftward shift in the voltage dependence of strongly desensitized TRPM4 channels after membrane excision into mM concentrations of Ca 2+ [12]. We therefore examined how transient depletion of endogenous PIP 2 by DrVSP activation affects the voltage dependence of whole-cell WT-and E7K-TRPM4 currents under the milder conditions where desensitization is much less evident, as in Figures 2 and 3. To make the time required to evaluate the voltage dependence as short as possible, we employed a dual-ramp protocol consisting of 1-s-long ascending and descending voltage ramps (−100-200 mV), between which a depolarizing step pulse to +120 mV was inserted (top panel in Figure 4A). The voltage ramps spanned the almost full range of the TRPM4 channel activation, and the 2-s depolarization was sufficient to maximally lower the membrane PIP 2 level (bottom panel in Figure 4A). In the absence of DrVSP, there was no discernible decrease of the current amplitude during depolarization to +120 mV and the following descending ramp in either WT-or E7K-mutant TRPM4 channels (data not shown). By this protocol, we could characterize the differences in voltage dependence before and after PIP 2 depletion between WT-and E7K-TRPM4 channels.
TRPM4 channels (Figure 3C), especially near the range of Kd values. Moreover, the recovery from the inhibition was significantly faster in E7K-than WT-TRPM4 channels ( Figure  3Da); the time constants of recovery were 13.71 ± 4.24 s and 17.56 ± 5.29 s for E7K and WT, respectively. These observations raise the idea that the E7K mutation may render the TRPM4 channel more tightly bound to endogenous PIP2, thereby stabilizing its activity. To minimize time-dependent changes due to desensitization, a relatively low concentration of Ca 2+ (1 μM) was included in the pipette to induce TRPM4 currents under the whole-cell conditions. The tree panels in the figure denote the simultaneously recorded whole-cell TRPM4 current (Im) and concomitant FRET ratio (FR) in response to depolarizing pulses of incremental duration (300-2000 ms; Vm). Prolongation of depolarizing pulses resulted in the progressive inhibition of whole-cell TRPM4 currents, the extent of which was greater for WT than E7K, despite the same degree of FR decrease (i.e., PIP2 depletion). (B) Duration-dependent inhibition of the whole-cell TRPM4 current by depolarization-induced PIP2 depletion for WT and E7K (open vs. filled circles). The durationdependent inhibition is defined by using the ratio of TRPM4 current amplitudes before (Ipre) and after (Ipost) each depolarization according to the formula: fractional inhibition = 1 − Ipost/Ipre, which is plotted against the pulse duration. FR change is also displayed together (dashed curve). (C) The relationships between endogenous PIP2 concentration and WT-or E7K-TRPM4 channel activity (normalized whole-cell current amplitude). Endogenous PIP2 concentration was estimated from the FR value after the previous study [16] according to the formula: where FRmax was determined from PIP5K-overexpressing cells as a 1.2-fold higher value than that evaluated from the control cells [16]. The apparent Kd values of PI(4,5)P2 binding to WT-TRPM4 channels and E7K-mutant TRPM4 channels determined by this method were 1.06 ± 0.05 μM and 0.97 ± 0.02 μM, respectively. *: p < 0.05 with ANOVA followed by Tukey's post hoc tests (n = 5). (D) Time courses of the whole-cell TRPM4 currents recovering from depolarization-induced, DrVSPmediated inhibition for WT (red curve in a) and E7K (blue curve in a). The time constant of recovery (τ) was determined by mono-exponential fitting, which resulted in values of 17.56 ± 5.29 and 13.71 ± 4.24 s for WT and E7K, respectively (b). *: p < 0.05 with unpaired t-test, respectively (n = 5). (A) Graded depletion of endogenous PI(4,5)P 2 was attained by applying a set of depolarizing pulses of incremental duration (300-2000 ms; from −60 to 120 mV). To minimize time-dependent changes due to desensitization, a relatively low concentration of Ca 2+ (1 µM) was included in the pipette to induce TRPM4 currents under the whole-cell conditions. The tree panels in the figure denote the simultaneously recorded whole-cell TRPM4 current (I m ) and concomitant FRET ratio (FR) in response to depolarizing pulses of incremental duration (300-2000 ms; V m ). Prolongation of depolarizing pulses resulted in the progressive inhibition of whole-cell TRPM4 currents, the extent of which was greater for WT than E7K, despite the same degree of FR decrease (i.e., PIP 2 depletion). (B) Duration-dependent inhibition of the whole-cell TRPM4 current by depolarization-induced PIP 2 depletion for WT and E7K (open vs. filled circles). The duration-dependent inhibition is defined by using the ratio of TRPM4 current amplitudes before (I pre ) and after (I post ) each depolarization according to the formula: fractional inhibition = 1 − I post /I pre , which is plotted against the pulse duration. FR change is also displayed together (dashed curve). (C) The relationships between endogenous PIP 2 concentration and WTor E7K-TRPM4 channel activity (normalized whole-cell current amplitude). Endogenous PIP 2 concentration was estimated from the FR value after the previous study [16] according to the formula: where FR max was determined from PIP5K-overexpressing cells as a 1.2-fold higher value than that evaluated from the control cells [16]. The apparent K d values of PI(4,5)P 2 binding to WT-TRPM4 channels and E7K-mutant TRPM4 channels determined by this method were 1.06 ± 0.05 µM and 0.97 ± 0.02 µM, respectively. *: p < 0.05 with ANOVA followed by Tukey's post hoc tests (n = 5). (D) Time courses of the whole-cell TRPM4 currents recovering from depolarization-induced, DrVSP-mediated inhibition for WT (red curve in a) and E7K (blue curve in a). The time constant of recovery (τ) was determined by mono-exponential fitting, which resulted in values of 17.56 ± 5.29 and 13.71 ± 4.24 s for WT and E7K, respectively (b). *: p < 0.05 with unpaired t-test, respectively (n = 5).  before (open circles) and after (filled circles) DrVSP-mediated PIP2 depletion. The values o were calculated from the data as shown in A, and the Gmax was taken as the maximal G val mV of the ascending ramp. After PIP2 depletion, the voltage dependence of the WT-TRPM4 was remarkably rightward shifted. In contrast, this shift was only marginal for E7K muta cially around the resting membrane potential. After fitting of these G-V data to the Boltzma tion: G/Gmax = G0/Gmax + (Gmax − G0)/(1 + exp((V0.5 − Vm)/s) (Vm, V0.5, s: membrane potential, ha mal activation voltage, slope factor, respectively), the V0.5 values before and after DrVSP ac 62.07 ± 7.32 and 83.31 ± 9.53 vs. 59.63 ± 10.79 and 219.84 ± 75.37; and the s values before a DrVSP activation (mV): 58.79 ± 11.46 mV and 70.45 ± 12.69 mV vs. 60.04 ± 18.52 mV and 26.29, were obtained for the E7K-mutant and WT-TRPM4 channels, respectively (n = 8). N in both WT and E7K, there appeared to be a substantial non-voltage-dependent com (G0/Gmax) at very negative potentials, which may reflect incomplete PIP2 depletion by a 2-s ization to +120 mV.

N-Terminal Polypeptides Inhibit TRPM4 Channel Activity
The hitherto-mentioned results indicate that the 'E7K' sequence in the N-te may be critical for enhanced PIP2 affinity and activity of the mutant TRPM4 chan substantiate this idea more directly, we created N-terminal polypeptides contain (EKE 5-7 ) or E7K (EKK 5-7 ) sequences and compared their inhibitory effects on TRPM nel activity. An empty vector (control) or that containing the first 100 N-termina acid residues of either WT or E7K was co-expressed with the full-length WT-TRPM nel into the HEK293 cells. In the whole-cell configuration at a holding potential of − intracellular perfusion of 1 μM Ca 2+ via a patch electrode-elicited robust TRPM4 was observed (not shown). The co-expression of either WT-or E7K polypeptide pipette significantly inhibited these currents, in particular around the resting me after (filled circles) DrVSP-mediated PIP 2 depletion. The values of G/G max were calculated from the data as shown in A, and the G max was taken as the maximal G value at 200 mV of the ascending ramp. After PIP 2 depletion, the voltage dependence of the WT-TRPM4 channel was remarkably rightward shifted. In contrast, this shift was only marginal for E7K mutant, especially around the resting membrane potential. After fitting of these G-V data to the Boltzmann equation: G/G max = G 0 /G max + (G max − G 0 )/(1 + exp((V 0.5 − V m )/s) (V m , V 0.5 , s: membrane potential, half-maximal activation voltage, slope factor, respectively), the V 0.5 values before and after DrVSP activation: 62.07 ± 7.32 and 83.31 ± 9.53 vs. 59.63 ± 10.79 and 219.84 ± 75.37; and the s values before and after DrVSP activation (mV): 58.79 ± 11.46 mV and 70.45 ± 12.69 mV vs. 60.04 ± 18.52 mV and 101.58 ± 26.29, were obtained for the E7K-mutant and WT-TRPM4 channels, respectively (n = 8). Note that in both WT and E7K, there appeared to be a substantial non-voltage-dependent component (G 0 /G max ) at very negative potentials, which may reflect incomplete PIP 2 depletion by a 2-s depolarization to +120 mV.
As illustrated in Figure 4B, after DrVSP-mediated PIP 2 depletion, the shift of halfmaximal activation voltage (V 0.5 ) was much smaller in E7K than WT, particularly around the physiological range of membrane potential (below +50 mV): V 0.5 values before and after VSP activation were 62.07 ± 7.32 and 83.31 ± 9.53 vs. 59.63 ± 10.79 and 219.84 ± 75.37 for E7K and WT, respectively (n = 8). Consequently, in the negative membrane potential range, WT channels were strongly suppressed after VSP-mediated PIP 2 depletion, whereas E7K channels were less affected ( Figure 4A). These results indicate that the E7K mutant may remain functional even after vigorous depletion of endogenous PIP 2 , possibly because of its tight binding to PIP 2 . Being consistent with this possibility, an N-terminal polypeptide carrying the E7K mutation tended to show a more potent binding to PI(4,5)P 2 than that of WT ( Figure S1). These results provide an interesting explanation that tighter binding to (or interaction with) endogenous PIP 2 may strengthen the voltage-dependent activation of the E7K-TRPM4 channel.
peptides significantly reduced the density of the whole-cell inward TRPM4 current, the E7K polypeptide was more effective than the WT polypeptide (e.g., at −100 mV, −3.23 ± 0.60 and −1.92 ± 0.43 pA/pF with WT-and E7K polypeptides, respectively, vs. −4.65 ± 0.78 pA/pF with an empty vector alone). Consistent with this observation, simultaneous application of shorter synthetic N-terminal polypeptides containing the WT and E7K sequences suppressed single-TRPM4 channel activities induced by 100 μM Ca 2+ (top and middle panels in Figure 5B), but the degree of the inhibition was greater with E7K-than WT polypeptides (0.56 ± 0.02 vs. 0.31 ± 0.03, respectively, n = 5) (bottom graph in Figure 5B). These results could be interpreted to mean that the 'E7K' sequence in the polypeptide more effectively hinders the N-terminal domain of TRPM4 channel protein from interacting with membrane PIP2. , or an empty vector alone (a). TRPM4 current was induced by 1 μM Ca 2+ under the whole-cell conditions. The densities of inward and outward TRPM4 currents were compared at membrane potentials of −120, −60, 100, and 200 mV (b). The E7K polypeptide was more effective than WT polypeptide at inhibiting whole-cell TRPM4 currents. *: p < 0.05 with ANOVA followed by Tukey's post hoc tests (n = 7). Only the pairs of columns that show statistically significant differences are shown. (B) Direct application of short synthetic N-terminal polypeptides containing the 'WT' and 'E7K' sequences suppressed single-TRPM4 channel activities induced by 100 μM Ca 2+ at −60 mV. Representative data from two inside-out patches showing greater inhibition by E7K-than WT polypeptides. *: p < 0.05 with unpaired t-test, respectively (n = 5).

Discussion
Membrane phospholipids serve as the structural foundation for intracellular signaling, which is most commonly initiated by receptor activation to alter a variety of ion channel/transporter activities. The PIP2 turnover is tightly linked to and capable of influencing the electrical properties of cardiomyocyte. Thus, altered PIP2 response of certain cardiac ion channels could cause arrhythmia. The densities of inward and outward TRPM4 currents were compared at membrane potentials of −120, −60, 100, and 200 mV (b). The E7K polypeptide was more effective than WT polypeptide at inhibiting whole-cell TRPM4 currents. *: p < 0.05 with ANOVA followed by Tukey's post hoc tests (n = 7). Only the pairs of columns that show statistically significant differences are shown. (B) Direct application of short synthetic N-terminal polypeptides containing the 'WT' and 'E7K' sequences suppressed single-TRPM4 channel activities induced by 100 µM Ca 2+ at −60 mV. Representative data from two inside-out patches showing greater inhibition by E7K-than WT polypeptides. *: p < 0.05 with unpaired t-test, respectively (n = 5).

N-Terminal Polypeptides Inhibit TRPM4 Channel Activity
The hitherto-mentioned results indicate that the 'E7K' sequence in the N-terminus may be critical for enhanced PIP 2 affinity and activity of the mutant TRPM4 channel. To substantiate this idea more directly, we created N-terminal polypeptides containing WT (EKE 5-7 ) or E7K (EKK 5-7 ) sequences and compared their inhibitory effects on TRPM4 channel activity. An empty vector (control) or that containing the first 100 N-terminal amino acid residues of either WT or E7K was co-expressed with the full-length WT-TRPM4 channel into the HEK293 cells. In the whole-cell configuration at a holding potential of −60 mV, intracellular perfusion of 1 µM Ca 2+ via a patch electrode-elicited robust TRPM4 current was observed (not shown). The co-expression of either WT-or E7K polypeptides in the pipette significantly inhibited these currents, in particular around the resting membrane potential (Figure 5Aa). As summarized in Figure 5Ab, although both WT-and E7K polypeptides significantly reduced the density of the whole-cell inward TRPM4 current, the E7K polypeptide was more effective than the WT polypeptide (e.g., at −100 mV, −3.23 ± 0.60 and −1.92 ± 0.43 pA/pF with WT-and E7K polypeptides, respectively, vs. −4.65 ± 0.78 pA/pF with an empty vector alone). Consistent with this observation, simultaneous application of shorter synthetic N-terminal polypeptides containing the WT and E7K sequences suppressed single-TRPM4 channel activities induced by 100 µM Ca 2+ (top and middle panels in Figure 5B), but the degree of the inhibition was greater with E7K-than WT polypeptides (0.56 ± 0.02 vs. 0.31 ± 0.03, respectively, n = 5) (bottom graph in Figure 5B). These results could be interpreted to mean that the 'E7K' sequence in the polypeptide more effectively hinders the N-terminal domain of TRPM4 channel protein from interacting with membrane PIP 2 .

Discussion
Membrane phospholipids serve as the structural foundation for intracellular signaling, which is most commonly initiated by receptor activation to alter a variety of ion channel/transporter activities. The PIP 2 turnover is tightly linked to and capable of influencing the electrical properties of cardiomyocyte. Thus, altered PIP 2 response of certain cardiac ion channels could cause arrhythmia.
In this study, our findings show that the E7K mutation is associated with an enhanced interaction of TRPM4 channels with PIP 2 . This mutation has been identified in a few pedigrees of patients' families associated with progressive conduction blocks and sudden death. Altered SUMOylation for the E7K mutation is suggested to be a molecular mechanism accounting for increased cell-surface expression of the TRPM4 channel protein and its consequent activity, which did not accompany changes in its gating behavior [6]. However, our preliminary results hinted that in this TRPM4 mutant, single-channel gating is also affected [22]. This suggests the presence of an additional functional modification, in which altered PIP 2 interaction could be involved, since it has been reported to have a significant impact on the TRPM4 channel's behavior [12].
There are several basic residues assumed to contribute to TRPM4 channel-PIP 2 interaction. In addition to rigorous interaction between PIP 2 and the C-terminal residues of this channel [12], recent studies identified an N-terminal domain that interacts with PIP 2 to exert an important functional modification [23]. It has also been proven that PIP 2 interacts with CaM-binding domains on the distal N-terminus of the TRPM3 channel [24]. Moreover, based on crystal structure information [25], a recent study suggested that TRP channel function could be modulated through the N-and C-terminal interactions [26]. All these studies support the possibility that PIP 2 can interact with TRP channels via multiple domains/regions in a complex manner. In our experiments, PIP 2 -binding assay as well as functional analyses with synthetic PIP 2 analogs, DrVSP-mediated PIP 2 depletion, and N-terminal polypeptides demonstrated that a few charged amino acids on the distal Nterminus may also be essential in some way for the functional regulation of the TRPM4 channel via PIP 2 interaction.
To gain more physiological insights about how endogenous PIP 2 modulates TRPM4 channel activity, we utilized the capability of DrVSP to control the endogenous PIP 2 level. More specifically, since DrVSP induces a graded PIP 2 depletion by varying the strength of membrane depolarization without involvement of other second messengers that might modulate TRPM4 currents, this intervention, when combined with a PIP 2 -reporting FRET probe (i.e., the CFPmse-PHd/YFPmse-PHd pair), allowed us to quantitatively evaluate the TRPM4 channel interaction with endogenous PIP 2 level in situ. Using these new experimental approaches, a good correlation was observed between FRET changes induced by DrVSP activation and the extent of concomitant TRPM4 channel inhibition, and this enabled us to compare differences in the interaction with endogenous PIP 2 between WTand E7K-TRPM4 channels. In our study, DrVSP-mediated PIP 2 depletion almost instantaneously occurred upon strong depolarization, while the recovery from the depletion likely reflected the resynthesis of PIP 2 in the membrane, which required a longer time course.
DrVSP-mediated PIP 2 depletion induced a clear rightward shift in the voltage dependence of the WT-TRPM4 channel ( Figure 4B), while the shift was almost absent in the E7K mutant, especially around the resting membrane potential. This means that the E7K mutation rendered TRPM4 channel almost insensitive to vigorous PIP 2 depletion. The large difference in voltage dependency, in particular around the resting membrane potential, would support a greater arrhythmogenic risk of the E7K mutation in the TRPM4 channel, in addition to its enhanced activity due to increased cell-surface expression [6]. The apparent potency of PIP 2 to modulate TRPM4 channel activity estimated from the functional analyses in Figure 3 indicates that the E7K mutant interacts more closely with PIP 2 than WT, but they were only slightly different from each other (K d values: 0.97 vs. 1.06 µM), which corresponds to only~0.1RT difference in free energy for PIP 2 binding (or interaction). A similarly small difference in PIP 2 dependence was observed between the reactivation curves of desensitized WT and E7K mutant channels by diC 8 PIP 2 (K d : 2.40 vs. 1.44 µM, respectively; equivalent to a free energy difference of~0.5RT; Figure 1C). These small differences in PIP 2 dependence appear insufficient to account for the largely discrepant PIP 2 dependence of voltage-dependent activation curves observed for the WT and E7K channels (i.e., the shifts of the curves by PIP 2 depletion were 21 and 160 mV, respectively; Figure 4B). In principle, the closed-to-open transition of voltage-dependent channels can be governed by two terms of free energy change, i.e., the voltage-independent conformational energy increase that occurs in the absence of membrane potential and the voltage-dependent energy increase or electrical energy increase proportionate to the movement of gating charges [27]. However, the large values of the slope factor of 60-100mV, which did not differ much between the WT and E7K mutant or change before and after DrVSP-mediated PIP 2 depletion (see the Figure 4 legend), indicate that the effective number of gating charges involved in the voltage-dependent activation of the TRPM4 channel would be far less than one, which is much smaller than those of purely voltage-dependent channels [27]. Therefore, the observed shift of the voltage-dependent activation curve for WT-TRPM4 channels upon PIP 2 depletion (Figure 4) would most likely reflect the voltageindependent conformational energy decrease resulting from the release of PIP 2 -mediated 'steric interactions' within the channel complex which are essential for its activity [28]. These interactions would be strongly tightened in the E7K mutant, and this may be the reason why vigorous, but not complete, PIP 2 depletion by DrVSP activation resulted in only a marginal decrease of E7K channel activity. At present, however, it remains entirely unclear what structural/mechanistic basis is involved therein, since the only known atomic structure of the TRPM4 channel is its 'closed' configuration, which lacks the detailed information about the most distal part of the N-terminus [29].
Finally, in order to seek the possible physiological/pathophysiological implications of the present findings, we performed single-cell action potential (AP) simulations with complex Ca 2+ dynamics for WT-and E7K-mutant TRPM4 channels. As performed previously, we adopted the Luo-Rudy 2000 ventricular AP model incorporated with TRPM4-gating kinetics [19]. Alterations in the rate constants of opening and closing were estimated from those obtained in the absence and presence of 5µM diC 8 PIP 2 ( Figure S2). The results of the simulation indicated that PIP 2 depletion prevents the arrhythmic changes in AP that would otherwise occur when TRPM4 protein expression and/or channel activity is pathologically upregulated 5-to 6-fold. In stark contrast, the presence of an E7K mutation almost eliminates this preventive effect of PIP 2 depletion, and renders the channel much more excitable. During sustained neurohormonal stresses in which cardiomyocytes progressively undergo remodeling changes, the expression of the TRPM4 channel is expected to dramatically increase [30], and concomitantly, the PIP 2 metabolism would be largely compromised [11]. Therefore, it is conceivable that the mechanism described above may help to reduce the arrhythmogenic risk as a negative feedback to suppress the excessive activation of TRPM4 channels, which is presumably disrupted by the E7K mutation. Moreover, it is tempting to speculate that during the cardiac remodeling processes, the overactivation of TRPM4 could occur in the different regions of the heart including the conduction system (e.g., Purkinje fibers), where an abnormal depolarizing shift of the resting membrane potential due to excessive TRPM4 channel activity may not only slow or halt AP conduction as the result of increased inactivation of voltage-dependent Na + channels, but might also cause fibrotic degeneration of Purkinje fibers that could lead to permanent conduction failures (see the Introduction Section). Further work is needed to verify this interpretation.
In summary, the present study revealed the physiological significance of TRPM4-PIP 2 interaction to maintain channel activity. By using FRET-based PIP 2 sensors in combination with voltage-sensing phosphatase (DrVSP) and patch clamping, we clearly demonstrated that the activity of the WT-TRPM4 channel is positively correlated with the endogenous PIP 2 level, which is abnormally strengthened in the E7K mutant through modulation of voltage-dependent gating. Such altered PIP 2 interaction may provide a novel pathogenic mechanism underlying TRPM4 channelopathies including arrhythmias.
Supplementary Materials: The following are available online at https://www.mdpi.com/article/10 .3390/cells10050983/s1, Figure S1: Phosphoinositide-binding of n-terminal peptides from WT and E7K mutant assessed by dot-blot assays. Figure S2: E7K mutation abrogates the preventive effect of a PIP2 decrease against arrhythmogenicity because of excessive TRPM4 activity.  Institutional Review Board Statement: Not applicable.

Informed Consent Statement: Not applicable.
Data Availability Statement: The data described in this paper are available on request.