Engineering an enhanced voltage-sensing phosphatase

The following cDNAs were used for expression in culture cells (HEK293T): Dr-VSP in pIRES2-EGFP (Hossain et al., 2008); mouse-Kir2.1 (Kir2.1) in pCXN2 (kindly provided by Y. Kubo, National Institute for Physiological Sciences, Okazaki, Japan); KCNQ2-YFP and KCNQ3-YFP (kindly provided by B. Hille, University of Washington, Seattle, WA); and human TRPC6 (available from GenBank under accession no. NM_004621) in pcDNA3 provided by T. Hofmann and M. Schaefer (Institut für Pharmakologie und Toxikologie, Würzburg, Germany).

Point mutations at Leu-223 in Dr-VSP were generated using a Quik Change Site-Directed Mutagenesis kit (Stratagene) or PrimeSTAR Mutagenesis Basal kit (Takara Bio). To produce mCherry-tagged constructs, the cDNA for Dr-VSP was amplified by PCR and ligated into a customized pcDNA3.1(–) (Invitrogen) expression vector, which yielded mCherry fused at the C-terminal end of the target protein (Dr-VSP^mChe). Chimeric proteins composed of Ci-VSP (M1-G103) and Dr-VSP (Q44-P511; CiDr-VSP) were generated through two-step PCR and ligated into the customized pcDNA3.1(–) vector.

The cDNAs for WT or mutant Dr-VSP were cotransfected with Kir2.1, KCNQ2/3, or TRPC6 into HEK293T cells using PEI Max (Polysciences, Inc.) or SuperFect (Qiagen), and Opti-MEM (Invitrogen). The cells were cultured for 6 h in Dulbecco’s modified Eagle’s medium (Invitrogen) supplemented with 10% FBS (GIBCO BRL) after mixing with transfection reagent. The medium was then refreshed. Electrophysiological recordings were performed 2–8 h after reseeding the transfected cells onto poly-L-lysine–coated glass-bottom dishes or coverslips (Matsunami Glass Ind.), which corresponds to 18–24 h after transfection. For recording, a dish containing HEK293T cells was placed on the stage of an inverted fluorescence microscope (IX72 or IX73, Olympus), and cells showing the fluorescence signal from EGFP (pIRES2-EGFP) or Dr-VSP^mChe were recorded. Macroscopic currents were recorded in either the whole-cell clamp or on-cell patch clamp configuration using an Axopatch-200B amplifier (Molecular Devices) or EPC800USB (HEKA).

To measure macroscopic Kir2.1 currents in the on-cell patch clamp configuration, the bath solution contained 100 mM KCl, 1 mM MgCl₂, 1 mM CaCl₂, and 20 mM HEPES, pH 7.2, while the pipette solution contained 10 mM KCl, 90 mM NaCl, 3 mM MgCl₂, 1 mM EGTA, and 20 mM HEPES, pH 7.2. To record Kir2.1 or KCNQ2/3 currents in the whole-cell mode, the bath solution contained 10 mM KCl, 90 mM NaCl, 1 mM MgCl₂, 1 mM CaCl₂, and 20 mM HEPES, pH 7.2, while the pipette solution contained 100 mM KCl, 3 mM MgCl₂, 1 mM EGTA, 2 mM ATP-Na₂, and 20 mM HEPES, pH 7.2. A stock solution of 100 mM ATP-Na₂ in 20 mM HEPES was mixed with the pipette solution before use. To measure sensing currents, we used solution containing 100 mM NMDG, 3 mM MgCl₂, 1 mM EGTA, and 180 mM HEPES, pH 7.0, on both sides of the cell membrane.

To measure TRPC6 currents, the standard external solution contained 140 mM NaCl, 5 mM KCl, 1 mM CaCl₂, 1.2 mM MgCl₂, 10 mM HEPES, and ∼10 mM glucose (pH 7.4, adjusted with Tris base; 300 mOsm, adjusted with glucose) and was always perfused. The pipette solution contained 120 mM Cs aspartate, 20 mM CsCl, 2 mM MgCl₂, 5 mM EGTA, 1.5 mM CaCl₂, 10 mM HEPES, 2 mM ATP-Na₂, 0.1 mM GTP, and ∼10 mM glucose (pH 7.2, adjusted with Tris base; 290–295 mOsm, adjusted with glucose). The diacylglycerol analogue 1-oleoyl-2-acetyl-sn-glycerol (OAG; Sigma-Aldrich), which is a TRPC6 activator, was diluted in the standard external solution from its stock concentration (50 mM in DMSO). TRPC6 currents were activated by perfusion of OAG. They reached the maximum amplitude in ∼30 s after application of OAG, and then the amplitude gradually decreased for several minutes (Itsuki et al., 2012). To verify monovalent cationic currents, the external solution was replaced with NMDG solution (150 mM NMDG chloride, 10 mM HEPES, and 1 mM CaCl₂, pH 7.4, adjusted with HCl) at the end of each stimulus. Stimulation, data acquisition, and analysis were performed on a computer using a Digidata 1200, 1440A, or 1550B AD/DA converter and pClamp 9.0, 10.3, or 10.7 (Molecular Devices). The pipette resistance in the solution was 2–3 MΩ for measuring ionic currents or 7–12 MΩ for measuring sensing currents. The recorded currents were low-pass filtered at 5 kHz using a four-pole Bessel filter circuit built into the amplifier, and the sampling frequency was 10–20 kHz. Current traces were digitally filtered at 1 kHz. All recordings were made at room temperature (22–26°C).

Cell imaging was performed using an electron multiplying charge-coupled device camera (ImageM C9100-13, Hamamatsu Photonics) mounted on an inverted epifluorescence microscope equipped with a 60× oil immersion objective lens (Plan ApoN, Olympus), a stable xenon lamp (U-LH75XEAPO, Olympus), and a filter set that included a BP542 (excitation filter), BP620 (emission filter), and DM570 (dichroic mirror; Olympus) along with the patch clamp setup. Image acquisition was controlled by a computer using MetaMorph (Molecular Devices). We obtained phase contrast (exposure time, 100 ms) and fluorescence images (exposure time, 1,000 ms) just before the final approach of the glass pipette to a target cell. All phase contrast images included the target cell and pipette edge (data not shown).

We used ImageJ software (National Institutes of Health) to quantify the fluorescence intensity of Dr-VSP^mChe in images acquired as described above. We first calculated the average intensity of whole cell area from the obtained images. We then drew an arbitrary line on the cell, and the fluorescence intensities along the line were measured and plotted on a graph, which enabled us to determine the peak intensity at the cell contour (Fig. 4). Background fluorescence intensities were subtracted based on the average fluorescence intensity in the void area.

Confocal imaging was performed using the confocal microscope (LSM710, Carl Zeiss) equipped with 63× oil immersion objective lens (with numerical aperture 1.2, Carl Zeiss). The sample was excited with the 561-nm laser and the signal was detected using a 575–610-nm bandpass filter. Image acquisition was controlled by a computer using ZEN (Carl Zeiss). An image was obtained from a single focal plane.

Data were analyzed using Excel (Microsoft), ImageJ or Fiji (National Institutes of Health), Clampfit (Molecular Devices), MATLAB (MathWorks), Prism (GraphPad Software), and Igor Pro (WaveMetrics) software. Data are presented as means ± SD. Statistical significance was assessed using two-tailed Student’s t tests or two-way ANOVA.

Dr-VSP is better expressed in mammalian cultured cells than in Xenopus laevis oocytes (Hossain et al., 2008; Okamura et al., 2009). We previously reported that the voltage-dependent phosphatase activity of Ci-VSP reflects the hydrophobicity of the structure on membrane–protein interface in the phosphatase domain (Leu-284 and Phe-285), which protrudes toward the membrane. We call this structure the hydrophobic spine (Kawanabe et al., 2018). Increasing the hydrophobicity of the spine in Ci-VSP through amino acid replacement increased the voltage-dependent enzyme activity. To test whether a similar change takes place upon amino acid substitution in the corresponding region of Dr-VSP, the WT enzyme or a mutant in which Leu-223 or Tyr-224 in the hydrophobic spine was replaced with Phe or Gln were heterologously expressed in HEK293T cells. An inward rectifying K⁺ channel (mouse Kir2.1) was used for the read out of the voltage-dependent phosphatase activity toward PI(4,5)P₂ in the plasma membrane, as previously established (Murata and Okamura, 2007; Sakata and Okamura, 2014; Rjasanow et al., 2015; Tang et al., 2015).

Fig. 1 shows representative current traces from cells coexpressing Dr-VSP WT or a hydrophobic spine mutant (L223F or L223Q) with Kir2.1 recorded in the on-cell patch clamp configuration. The pulse protocol was composed of two parts (Fig. 1 A): (1) a 20-ms step pulse to −100 mV to determine the steady-state Kir current amplitude (test pulse), and (2) a 300-ms depolarizing step pulse to 100 mV to activate Dr-VSP. A large Kir current was detected in the first episode during the test pulse to −100 mV (Fig. 1 B). This was followed by a gradual decrease in the Kir current amplitude with repeated stimulations. When the normalized Kir currents measured at the end of the 20-ms test pulse were plotted against the accumulated depolarization time (e.g., first, 0 ms; second, 300 ms; Fig. 1 C), the kinetics of the decrease in the Kir current reflected the voltage-dependent phosphatase activity. The rate constants of single exponential decay of Kir current were compared in Fig. 1 D. The L223F mutant produced a more rapid Kir current decrease than did WT Dr-VSP (Fig. 1, B–D; and Table 1). By contrast, L223Q, in which a hydrophilic amino acid was introduced into the hydrophobic spine, produced a slower Kir current decrease, indicating that the voltage-dependent phosphatase activity is diminished in the mutant (Fig. 1, B–D). Another hydrophilic mutant, Y224Q, was also tested, and it showed less or no phosphatase activity (Fig. 1 D). These results are consistent with an earlier report on Ci-VSP (Kawanabe et al., 2018).

VSP has also been studied with another voltage-gated K⁺ channel, KCNQ2/3 (Kv7.2/7.3; Rjasanow et al., 2015; Hossain et al., 2008; Falkenburger et al., 2010). KCNQ2/3 binds PI(4,5)P₂ with weaker affinity than does Kir2.1, and the dynamic range of the sensitivity differs between the two channels (Rjasanow et al., 2015). Fig. 2 shows representative outward KCNQ2/3 current traces recorded from cells expressing Dr-VSP WT or a hydrophobic spine mutant. The pulse protocol consisted of a test pulse (0 mV, 300 ms) and a depolarization step (75 mV, 300 ms, for activating VSP; Fig. 2 A). A slowly activating outward current was observed in the first trace during a test pulse to 0 mV, and the current amplitudes decreased in the subsequent traces with all three constructs (Dr-VSP WT, L223F, and L223Q). When the time constants for the decrease in KCNQ2/3 currents were estimated by fitting with a single exponential function, L223F induced a faster decrease in the KCNQ2/3 current than did WT Dr-VSP (Fig. 2, B–D; and Table 1). On the other hand, the current decrease exhibited slower kinetics in cells expressing L223Q (Fig. 2, B–D). These results are consistent with those obtained using Kir2.1 as the PI(4,5)P₂ reporter.

We have shown that Dr-VSP with a Phe mutation within the hydrophobic spine has higher voltage-dependent phosphatase activity than the WT enzyme. However, the apparent enzymatic activity of VSP will also depend on the level of its surface expression (Murata and Okamura, 2007). In Xenopus oocytes, the expression level can be controlled by regulating the amount of injected RNA (Rayaprolu et al., 2018). On the other hand, the protein expression level in transfected cultured cells often varies from cell to cell dependent on the copy number of introduced plasmids.

Therefore, we detected the cell surface fluorescence signal from a fluorescent protein (mCherry) fused to Dr-VSP (Dr-VSP^mChe). After transfection into HEK293T cells, we verified the voltage-dependent phosphatase activity of Dr-VSP^mChe by measuring the time course of the decrease in Kir2.1 currents by whole-cell patch recording (Fig. 3 A), indicating that fusion of mCherry does not affect the intrinsic phosphatase activity of Dr-VSP.

We noticed that the phosphatase activities in Dr-VSP^mChe showed a large variance and cells with bright fluorescence often showed little or no phosphatase activity, as detected through activation of coexpressed K⁺ channels. We suspect that overexpression of Dr-VSP^mChe may have had a toxic effect on these cells, possibly due to mistargeting of the overexpressed Dr-VSP^mChe to intracellular compartments. Confocal microscopic examination showed that the fluorescence signal from Dr-VSP^mChe not only was present at the cell surface but was also widely distributed throughout the intracellular compartments (Fig. 3 C, left, mCherry).

To facilitate cell surface expression, the N-terminal cytoplasmic region of Ci-VSP (M1-G103), which is known to facilitate the enzyme’s membrane targeting (Tsutsui et al., 2013), was fused to the main body of Dr-VSP (Q44-P511; Fig. 3 D). Most cells transfected with this chimeric protein exhibited a more discrete distribution that most likely corresponded to the plasma membrane, suggesting CiDr-VSP^mChe is more selectively localized to the cell membrane than Dr-VSP^mChe (Fig. 3 C).

To confirm whether this chimeric version of Dr-VSP harboring the N-terminal region of Ci-VSP is more highly localized to the plasma membrane than Dr-VSP, we compared the fluorescence intensity and the sensing charge from the same cells. Sensing (or gating) charges are known to provide a reliable quantitative estimate of functional voltage-gated ion channels on cell membranes (Hille, 2001) and VSP (Murata and Okamura, 2007). We first acquired a fluorescence image of Dr-VSP^mChe-expressing cells using an electron multiplying charge-coupled device camera mounted on an epifluorescence microscope within the patch clamp setup, and then made whole-cell recordings to measure the sensing currents. As shown in the representative images of Dr-VSP^mChe and CiDr-VSP^mChe (Fig. 4 A), we evaluated two signals in the fluorescence images: the average intensity over the whole cell area and the intensity at the cell contour (Fig. 4 B). The fluorescence intensity at the contour was estimated by measuring the peak intensity at the cell “edge” along a white line on cell fluorescence images (Fig. 4 A). We detected no difference of the fluorescence magnitude over the whole cell area or cell contour between Dr-VSP^mChe and CiDr-VSP^mChe (Fig. 4 B).

Dr-VSP sensing currents were measured in whole-cell patch recordings with the NMDG-based intracellular and extracellular solution appropriate for sensing currents (Fig. 4 C). The off-sensing charges (Qoff) were calculated by integrating the sensing currents upon repolarization step and plotted against the depolarized membrane voltage of test pulse (charge-voltage [Q-V] relationship; Fig. 4 D). CiDr-VSP^mChe exhibited greater Qoff than Dr-VSP^mChe with no significant change of its voltage dependency (Table 2), which suggests that CiDr-VSP^mChe is more localized to the plasma membrane than Dr-VSP^mChe. Q-V curves showed that the magnitude of the Qoff significantly differs between Dr-VSP^mChe and CiDr-VSP^mChe (P = 0.05 by repeated-measures two-way ANOVA; see Fig. 4 D legend for details). Q-V relationships of four constructs (Dr-VSP^mChe WT, L223F and CiDr-VSP^mChe WT, L223F) are shown in Fig. 4 D (lower panel). Magnitudes of Qoff in Dr-VSP^mChe L223F and CiDr-VSP^mChe L223F show similar difference as in the WT pair, verifying that fusion of the N-terminal region of Ci-VSP to Dr-VSP enhances cell surface expression. In either case, with or without N-terminal Ci-VSP sequence, the Qoff curves for WT and L223F were superimposed on each other, suggesting that voltage dependence was not significantly altered by mutation in the hydrophobic spine (Table 2).

A plot of the fluorescence intensity of mCherry in the cell contour against Qoff (charge-fluorescence [Q-F] relationship; Fig. 4 E) shows that the linear relationship between Qoff and fluorescence is steeper for CiDr-VSP^mChe than for Dr-VSP^mChe (slope, 1.78 × 10⁻⁴ in Dr-VSP^mChe; 0.95 × 10⁻⁴ in CiDr-VSP^mChe), suggesting that the Dr-VSP^mChe fluorescence of the cell contour (Fig. 4, A and B) estimated based on epifluorescence microscopy includes any signal beneath the cell membrane. Qoff was measured at the holding potential after depolarizing to 130 mV, where the magnitude of sensing charge does not attain a saturated level, since we could not obtain the data due to unstable membrane at the higher voltage. Thus, detection of cell surface expression level of Dr-VSP^mChe by epifluorescence microscopy provides a less reliable estimate than quantification by measuring sensing current. Most importantly, these results are consistent with the idea of more selective cell membrane localization of CiDr-VSP^mChe than in Dr-VSP^mChe.

The L223F mutation within the hydrophobic spine of Dr-VSP enhanced its phosphatase activity. In addition, creation of the CiDr-VSP^mChe chimera enhanced its selective accumulation in the cell membrane. These two features were combined in CiDr-VSP^mChe L223F (eVSP). The enzyme activity of CiDr-VSP^mChe L223F (eVSP) was examined using Kir2.1 channel currents as the readout. Fig. 5 shows a comparison of the voltage-dependent enzyme activity among four versions, Dr-VSP^mChe WT, Dr-VSP^mChe L223F, CiDr-VSP^mChe WT, and CiDr-VSP^mChe L223F (eVSP), as reflected by the Kir2.1 channel activity (Fig. 5 A).

Among the four constructs, CiDr-VSP^mChe L223F (eVSP) mediated the most rapid current decrease when these were examined with repeated depolarizing pulses to 100 mV (Fig. 5, A–C; and Table 1). CiDr-VSP^mChe WT showed similar phosphatase activity of the Dr-VSP^mChe WT (P = 0.46). However, when the phosphatase activity of both constructs was measured at 75 mV (Fig. 5 D), CiDr-VSP^mChe WT (1.8 ± 1.6 s^-1) showed slightly higher activity than Dr-VSP^mChe WT (0.6 ± 1.0 s^-1, P = 0.10). CiDr-VSP^mChe L223F (eVSP) showed significantly higher phosphatase activity than Dr-VSP^mChe L223F (P = 0.03; Fig. 5 C). These results support the idea that increase of cell surface expression by the addition of the N-terminal region of Ci-VSP contributed to increase of the activity of voltage-dependent enzyme. On the other hand, there is also increase in phosphatase activity measured at 75 or 100 mV from Dr-VSP^mChe WT to L223F mutant (P = 0.01 or 0.13), and from CiDr-VSP^mChe WT to its L223F mutant (eVSP; P = 0.03 or 0.007; Fig. 5 C). We also performed two control experiments with CiDr-VSP^mChe L223F/C302S (C302S is a defective mutation of Dr-VSP’s enzyme activity; Hossain et al., 2008) or Kir2.1 alone. Both showed no current decay during application of the same pulse protocol of other constructs (Fig. 5, A–C). Measurements at different depolarizing membrane potentials (Fig. 5 D) showed that the four active constructs produced more rapid decay of Kir2.1 channel currents as the depolarizing membrane potential becomes higher (Fig. 5 D) and that decay kinetics seem almost saturated at 100 mV in each construct. Fold increases of voltage-dependent enzyme activity estimated by the Kir2.1 channel activities among different constructs (Table 1) indicate that facilitation of selective cell surface expression as well as mutation of L223F both contributed to enhancement of the activity of voltage-dependent decrease of PI(4,5)P₂ by eVSP.

We tried to validate the usefulness of eVSP as a tool using TRPC6, which is also sensitive to PI(4,5)P₂ (Imai et al., 2012; Itsuki et al., 2012). In contrast to Kir2.1, with TRPC6, the readout channel activity can be monitored during a single long depolarizing pulse that simultaneously activates the channel and VSP, thereby making repetition of depolarizing pulses for stimulation of VSP unnecessary. The traces in Fig. 6 show representative outward and inward Ca²⁺ currents through TRPC6 with CiDr-VSP^mChe WT and L223F (eVSP) in the presence of the TRPC6 activator OAG. A test pulse was stepped to 100 mV (Fig. 6 A). At the holding potential of −50 mV, small stable inward TRPC6 currents were observed, which were not present in the absence of OAG. During depolarization, large outward currents were observed. Then, a gradual current decrease was observed with CiDr-VSP^mChe WT, whereas a more rapid current decrease was observed with CiDr-VSP^mChe L223F (eVSP; Fig. 6 B), indicating that PI(4,5)P₂ depletion by eVSP is faster. We estimated “time lag” between the onset of the depolarization (+100 mV) and the onset of the current suppression (cross points of dotted lines), which reflects the phosphatase activity (Fig. 6, C and D; and Table 1). Upon repolarization, gradual current recovery was observed. Although recovery rates seemed to be different between the two constructs in traces of B, there was no statistical difference (time constant of recovery phase: CiDr-VSP^mChe WT, 2.6 ± 0.4 s [n = 6]; CiDr-VSP^mChe L223F [eVSP], 3.2 ± 0.6 s [n = 3], P = 0.58 by two-tailed Student’s t test). Cells coexpressing TRPC6 with the enzyme-defective mutant, C302S, of Dr-VSP (Hossain et al., 2008) did not exhibit depolarization-induced current decline (Fig. 6 B, bottom).

We have been able to enhance the activity of Dr-VSP by modifying the protein in two ways. The voltage-dependent phosphatase activity of Dr-VSP was enhanced by introducing an aromatic residue at the position of Leu-223 within the hydrophobic spine, a membrane-interacting region of the phosphatase domain. In addition, selective plasma membrane targeting of Dr-VSP was enhanced by fusion with the N-terminal region of Ci-VSP, which was previously shown to facilitate cell surface expression. The increased enzymatic activity of this modified version of Dr-VSP (CiDr-VSP^mChe L223F, eVSP) was verified based on the activities of coexpressed Kir2.1 and TRPC6 channels.

We assume that eVSP will serve as a better tool than the plain Dr-VSP based on two factors: better selective cell surface expression and higher phosphatase activity. eVSP showed an ∼2.8-fold increase of phosphatase activity over Dr-VSP as examined at 100 mV using the Kir2.1 as the readout (Fig. 5 C and Table 1). We were unable to determine how cell surface expression correlated with increase of phosphatase activity in HEK293T cells, since sensing charges and phosphatase activity were not able to be analyzed in the same cells in our experimental setup. Our trial of quantification of cell surface expression of Dr-VSP^mChe and its mutant by epifluorescence microscope was unsuccessful, since signals beneath the cell membrane were not completely separated. It is unlikely that apparent increase of voltage-dependent phosphatase activity of L223F mutation is through facilitated cell surface expression for two reasons. First, values of sensing charge, representing the number of cell surface VSP molecules, were not significantly different between eVSP and CiDr-VSP^mChe WT (Fig. 4). In addition, our previous study (Kawanabe et al., 2018) showed that Ci-VSP L284F, the construct with mutation of hydrophobic spine which corresponds to L223F mutation of Dr-VSP, did not show significant difference in cell surface expression from that of Ci-VSP WT. In the paper, detailed analysis of motions both of the VSD and the cytoplasmic enzyme region led to the conclusion that both phosphatase activity and coupling efficiency were facilitated in Ci-VSP L284F without showing a shift of voltage dependence of VSD motion. In the present study as well, sensing charges exhibited similar voltage dependence among the WT and L223F mutant (Table 2). Voltage dependence of enzyme activity of VSP is dictated by that of the VSD as long as the VSD is tightly coupled with the enzyme region (Sakata et al., 2011; Sakata and Okamura, 2014). Therefore, we assume that eVSP has higher maximum enzyme activity per molecule than WT Dr-VSP without shift of the voltage-enzyme relationship. More rigorous quantitative analysis such as analysis of raw results using the mathematical models (Sakata et al., 2011; Keum et al., 2016) may help to clarify in more detail how properties of eVSP are distinct from those of WT.

Many ion channels and transporters have been shown to be sensitive to PIs (Suh and Hille, 2005; Hille et al., 2015; Okamura et al., 2018). However, in cases where the binding affinity between the ion channel and PI is very high, the functional significance of the PI-channel interaction may be unknown. For example, the recently solved protein structure of the γ-aminobutyric acid A (GABA_A) receptor contains two molecules of PI(4,5)P₂ within the single channel complex (Laverty et al., 2019), but the functional significance of this PI(4,5)P₂ binding has yet to be determined (Mennerick et al., 2014), possibly due to the very high binding affinity and the inability to experimentally induce release of the PI(4,5)P₂ from the channel complex.

CiDr-VSP^mChe L223F (eVSP) could potentially serve as a useful tool in future studies of the function of PI(4,5)P₂ tightly bound to ion channels in heterologous mammalian cells. Experiments with various cell lines and primary cultures (e.g., muscle cells or neurons) will be necessary to further evaluate the usefulness of eVSP. In addition, VSP shows enzyme activity not only toward PI(4,5)P₂ but also for PI(3,4,5)P₃ and PI(3,4)P₂. Whether enzyme activity for PI(3,4,5)P₃ or PI(3,4)P₂ is also enhanced in eVSP remains an important question to be addressed in the future. Finally, we anticipate that further remodeling based on the molecular mechanisms of VSP will lead to further improvements in the properties of VSP for use as a molecular tool.

David A. Eisner served as editor.

We thank Dr. Yoshihiro Kubo for providing Kir2.1 plasmid, Dr. Bertil Hille for providing KCNQ2-YFP and KCNQ3-YFP plasmids, and Drs. Thomas Hofmann and Michael Schaefer for providing human TRPC6 plasmid. We also thank Dr. Yasuo Mori for encouragement and the Center for Medical Research and Education, Graduate School of Medicine, Osaka University for technical support.

This work was supported by Ministry of Education, Culture, Sports, Science and Technology Japan/Japan Society for the Promotion of Science KAKENHI grant nos. JP25253016 (Y. Okamura), JP15H05901 (Y. Okamura), JP19H03401 (Y. Okamura), and JP19K06585 (A. Kawanabe), and Japan Science and Technology Agency CREST grant no. JPMJCR14M3 (Y. Okamura).

The authors declare no competing financial interests.

Author contributions: A. Kawanabe, N. Mizutani, O.K. Polat, and T. Yonezawa performed experiments. A. Kawanabe, T. Kawai, and O.K. Polat analyzed the data. A. Kawanabe, M.X. Mori, and Y. Okamura designed the experiments. A. Kawanabe and Y. Okamura wrote the paper.