Nonsubstrate PI(4,5)P₂ interacts with the interdomain linker to control electrochemical coupling in voltage-sensing phosphatase (VSP)

Significance

PI(4,5)P₂ is a fundamental modulator of voltage-gated ion channels (VGICs). The PI(4,5)P₂ phosphatase, voltage-sensing phosphatase (VSP), in which a cytoplasmic catalytic region (CCR) is connected with a voltage sensor domain (VSD) similar to that of VGICs, is also regulated by PI(4,5)P₂. However, detailed physical interaction between VSP and PI(4,5)P₂ remains unclear. Here, we demonstrate that PI(4,5)P₂, which differs from the substrate one at the active site, interacts with residues in the interdomain linker, regulating electrochemical signal transduction through VSD-CCR coupling. This also points toward a regulatory mechanism of VSP on the plasma membrane through negative feedback loop, which may contribute to physiological functions of VSP such as motility and maturation of mouse sperm and nutrient absorption by teleost enterocyte.

Abstract

Voltage-sensing phosphatase (VSP) comprises a voltage sensor domain (VSD) and a cytoplasmic catalytic region (CCR), achieving a unique electrochemical signal conversion. Previous studies suggest that phosphatidylinositol 4,5-bisphosphate (PI(4,5)P₂), a membrane phospholipid known to be critical for activities of diverse voltage-gated ion channels, associates with a linker connecting the VSD with the CCR of VSP and regulates VSD-CCR coupling. However, the details of PI(4,5)P₂ interaction with the linker of VSP remain elusive. Here, we exploit advantage of sensitivity of a fluorescent unnatural amino acid, 3-(6-acetylnaphthalen-2-ylamino)-2-aminopropanoic acid (Anap), to changes in local environment to study interaction between PI(4,5)P₂ and the linker of Ciona intestinalis VSP (Ci-VSP). We found that a conserved tyrosine residue (Y255) as well as neighboring basic residues interacts with PI(4,5)P₂ and this interaction was maintained in G365A Ci-VSP mutant which lacks the substrate PI(4,5)P₂ at the active site and Ci-VSP/human phosphatase and tensin homolog (PTEN) chimera which does not dephosphorylate PI(4,5)P₂, indicating that the linker interacts with nonsubstrate, regulatory PI(4,5)P₂ outside the active site. Molecular dynamics simulations demonstrated that the linker formed stable interaction with PI(4,5)P₂ in the activated state. These findings indicate that regulation of coupling to an effector region downstream of the VSD through PI(4,5)P₂ binding to the linker is shared among voltage-dependent membrane proteins.

As one species of membrane proteins that respond to voltage changes across the plasma membrane by the voltage sensor domain (VSD) including voltage-gated ion channels (VGICs), voltage-sensing phosphatase (VSP) exhibits a distinctive electrochemical signal conversion by a tight coupling between a VSD and a cytoplasmic catalytic region (CCR) (1–3). The CCR consists of a phosphatase domain (PD) and a C2 domain (Fig. 1A), which shows high homology to a tumor suppressor enzyme, phosphatase and tensin homolog deleted on chromosome 10 (PTEN) (2). The interdomain linker connecting the VSD with the PD (VSD-PD linker) is divided into the N-terminal helical part and the following nonhelical part based on the predicted structure [M240 to S249 and Q250 to K257 of Ciona intestinalis VSP (Ci-VSP), respectively; Fig. 2A]. In response to membrane depolarization, VSP dephosphorylates a membrane phospholipid, mainly phosphatidylinositol 4,5-bisphosphate (PI(4,5)P₂) (4, 5). In the mouse spermatozoa, VSP is activated at the physiological membrane potential and this phosphatase activity creates a biased PI(4,5)P₂ distribution along the sperm flagellum, playing a critical role in regulation of its motility (6, 7). VSP is also expressed in zebrafish enterocytes and regulates endocytosis-dependent nutrient absorption (8), indicating a physiological importance of VSP in various tissues across animal species.

Fig. 1.

PI(4,5)P₂-dependent fluorescence component of Anap at K555 within the CCR of Ci-VSP. (A) Left: Structural model of Ci-VSP predicted using ColabFold (9). C_α atom of K555 is shown as a magenta sphere. Right: Ribbon diagram of Ci-VSP CCR with IP₃ bound in the active site C363 (PDB 3V0H) shown in lateral view. Side chains are shown as sticks for Y522 (green), K553 (orange), and K555 (magenta). Both K553 and K555 compose the Cα2 loop colored cyan. Dotted lines indicate hydrogen bonds. C_α atom of C363 is shown as a sphere (yellow) in both panels. (B) Comparison of fluorescence time courses of K555Anap upon depolarization to 160 mV of the long interval protocol in the absence (red, from SI Appendix, Fig. S2A) and the presence (blue, from SI Appendix, Fig. S2C) of WT Ci-VSP (Left) and the time constant (τ) of the fluorescence decay determined by fitting it with a single exponential function (Right). The fluorescence was detected using a 460- to 510-nm band-pass emission filter. Black dotted curves in the Left panel are single exponential fits. Bars are means ± SD; n = 5 and 4 for K555Anap without and with WT Ci-VSP, respectively; **P < 0.01, two-tailed unpaired t test.

Fig. 2.

PI(4,5)P₂-dependent structural rearrangements of Y255. (A) Left: Superimposition of structural models of human PTEN (hPTEN, light purple) and Ci-VSP (light orange) predicted using ColabFold (9). The N-terminal intracellular linker of Ci-VSP and the C-terminal tail of hPTEN were missed for clarity. Right: Close-up view of the putative PI(4,5)P₂-binding region of hPTEN and the VSD-PD linker of Ci-VSP surrounded with a black box in the Left panel. The conserved basic residues and tyrosine residue of hPTEN (K13, R14, and R15 and Y16) and Ci-VSP (K252, R253, and R254 and Y255) are shown as sticks and colored green and red, respectively (see also SI Appendix, Fig. S1). The Lower panel is sequence alignment of hPTEN and Ci-VSP. (B and C) Left: Representative fluorescence traces of Y255Anap from the same oocyte expressing Y255Anap Ci-VSP alone (B) and coexpressing with WT Ci-VSP (C) in response to step to 160 mV recorded using the long (red and blue, Long) and the short (black, Short) interval protocols were superimposed. Red (B) and blue (C) arrows indicate a change in the time course recorded with long interval. Cyan and magenta arrow heads indicate fluorescence components recorded right after depolarization (F_initial) and right after repolarization to −60 mV (F_repo), respectively. The fluorescence was detected using a 460- to 510-nm band-pass emission filter. Traces are averaged from five runs. Right: Changes in the amplitudes of F_peak − F_ss recorded with long (L) and short (S) intervals. F_peak − F_ss values were normalized to F_ss values. Bars are means ± SD; n = 9 and 6 for Y255Anap without and with WT Ci-VSP, respectively; *P < 0.05, **P < 0.01, two-tailed paired t test. Inset in B shows the definition of F_peak, F_initial, F_repo, and F_ss. The black line represents fluorescence level at −60 mV. (Scale bars: 200 ms, 1.0%.) (D) Comparison of fluorescence time courses of Y255Anap upon depolarization to 160 mV with long interval in the absence (red, from B) and the presence (blue, from C) of WT Ci-VSP (Left), and the time to reach F_peak from the peak of F_initial (t_peak) and τ of the fluorescence increase determined by fitting it with a single exponential function (Right). Black dotted curves in the Left panel are single exponential fits. Bars are means ± SD; n = 9 and 6 for Y255Anap without and with WT Ci-VSP, respectively; *P < 0.05, ***P < 0.001, two-tailed unpaired t test.

Molecular basis for VSD-CCR coupling of VSP has long been studied as one of the most important issues and it would also lead to our better understanding of detailed mechanisms underlying coupling between the VSD and the pore-gate domain in VGICs. Our recent study revealed that interaction between the C-terminal hydrophobic residues of the fourth transmembrane voltage-sensing helix, S4, and a hydrophobic membrane-bound structure within the PD (hydrophobic spine) plays an essential role in the coupling mechanism (3). Similar physical relationship of the hydrophobic residues within the N-terminal helix and the hydrophobic spine in PTEN (3) suggests that the VSD hijacks the innate regulatory mechanisms underlying phosphatase activity of PTEN. PI(4,5)P₂, which is an important regulator of a wide variety of ion channels and transporters (10–12), is shown to be an activator of PTEN based on the observation that the phosphatase activities toward phosphatidylinositol 3,4,5-trisphosphate (PI(3,4,5)P₃) and phosphatidylinositol 3,4-bisphosphate (PI(3,4)P₂) are potentiated in the presence of PI(4,5)P₂ (13). Substitutions of the conserved two consecutive positively charged residues (K13 and R14 of human PTEN; SI Appendix, Fig. S1) within the N-terminus with negatively charged or noncharged residues suppressed the effects (13–15), suggesting that the N-terminus of PTEN binds PI(4,5)P₂. Remarkable sequence and structural similarities between this putative PI(4,5)P₂-binding region of PTEN and the VSD-PD linker of VSP (SI Appendix, Fig. S1) raise a possibility that PI(4,5)P₂ modulates VSP. In fact, substitutions of the consecutive basic residues (K252, R253, and R254 of Ci-VSP; SI Appendix, Fig. S1) within the C-terminal part of the VSD-PD linker with a glutamine or an alanine changed VSD motion and induced coupling defects (16, 17), suggesting that PI(4,5)P₂ is an essential cofactor of VSD-CCR coupling associated with the VSD-PD linker. Molecular dynamics (MD) simulations showed that positively charged residues of the linker including R254 interact with PI(4,5)P₂ head groups (18). However, there is no direct evidence that the VSD-PD linker interacts with nonsubstrate PI(4,5)P₂ outside the active site and how the interaction achieves coupling in VSP. So far, no structural information of the full-length VSP protein containing the VSD-PD linker has been available despite recent cryo-electron microscopy (cryo-EM) structural analysis (19).

A fluorescent unnatural amino acid, 3-(6-acetylnaphthalen-2-ylamino)-2-aminopropanoic acid (Anap), is sensitive to changes in local environment of the incorporation site and voltage-dependent local structural rearrangements of residues in Ci-VSP have been detected by voltage clamp fluorometry (VCF) (20) with Anap (3, 19, 21–23). Anap incorporated at K555 near the enzyme active site showed a difference in the fluorescence time courses between normal and low levels of PI(4,5)P₂ in Xenopus oocytes (21), implying that Anap is a useful tool for reporting availability of PI(4,5)P₂ located close to its incorporation site. In the present study, to reveal interaction between the VSD-PD linker and PI(4,5)P₂, we thoroughly investigated fluorescence changes of Anap incorporated into residues within the linker of Ci-VSP by VCF. With optimized setup and voltage protocol, we found that the fluorescence time course of Anap incorporated at residues in the VSD-PD linker was also sensitive to availability of PI(4,5)P₂. Between normal and low level PI(4,5)P₂ conditions, a difference in the time courses suggests that the VSD-PD linker interacts with PI(4,5)P₂. Findings of Anap fluorescence recordings from VSP mutants with much weaker 5-phosphatase activity toward PI(4,5)P₂ compared to that of the wild-type (WT) VSP [G365A Ci-VSP (5) and Ci-VSPTEN16 (24)] as well as those of MD simulations showed that PI(4,5)P₂ interacts with the VSD-PD linker is not the same molecule as that for the substrate at the active site. Our findings suggest that nonsubstrate PI(4,5)P₂ acting on the VSD-PD linker plays a critical role in VSD-CCR coupling.

Results

Anap Incorporated at K555 in Ci-VSP Can Sense PI(4,5)P₂.

In our earlier study (21), we performed VCF on Xenopus oocytes expressing Ci-VSP harboring Anap in K555 within the Cα2 loop located at the C2 domain. The X-ray crystal structure of the CCR of Ci-VSP showed that the Cα2 loop has indirect interaction with the substrate at the active site (C363) through hydrogen bonds mediated by Y522 (25) (Fig. 1A). We found that an averaged trace of Anap fluorescence in response to repetitive membrane depolarization to 160 mV exhibited distinct time courses between the two protocols with distinct pulse intervals: An initial transient was observed with 60-s interval but not with a depolarizing prepulse prior to the fluorescence recording. These suggest that the change of fluorescence depends on the availability of PI(4,5)P₂. Using a higher sensitive microscope setup, in the present study we repeated this experiment at a better resolution and recapitulated this observation (long interval; SI Appendix, Fig. S2A, red trace). The initial transient was not observed in the same oocyte when a 500-ms depolarizing pulse to 160 mV was repeatedly applied with 0.7-s interval (short interval; SI Appendix, Fig. S2A, black trace). The amplitude of the initial transient (F_peak − F_ss) normalized to that at the steady state (F_ss) in the “long interval” was significantly larger than that in the “short interval” (SI Appendix, Fig. S2A). The level of available PI(4,5)P₂ on the plasma membrane is expected to gradually decrease during the depolarizing step for fluorescence recording (SI Appendix, Fig. S2 B, Upper). In the short interval protocol, 0.7-s interval is not sufficient for recovery of PI(4,5)P₂ to the original level, making the level of available PI(4,5)P₂ lower than that with the long interval protocol (SI Appendix, Fig. S2 B, Lower). Thus, the difference in the fluorescence time courses of K555Anap recorded between the two protocols with long and short intervals reflected the difference in PI(4,5)P₂ availability at the initiation of fluorescence recording.

To confirm that the time course of K555Anap fluorescence depends on the change of availability of PI(4,5)P₂, we recorded fluorescence changes of K555Anap with coexpression of the WT Ci-VSP in order to facilitate depletion of PI(4,5)P₂ more remarkably than in cells only expressing K555Anap Ci-VSP, which shows weaker phosphatase activity than the WT (21). The initial transient of Anap fluorescence was observed at depolarizing pulse upon the long interval protocol (SI Appendix, Fig. S2C, blue trace), and F_peak − F_ss was significantly decreased by using the short interval protocol in the same oocyte (SI Appendix, Fig. S2C, black trace). The time constant (τ) of the fluorescence decrease determined by fitting it with a single exponential function for K555Anap with WT Ci-VSP was significantly shorter than that for K555Anap alone (34.3 ± 10.1 ms versus 71.1 ± 14.3 ms; Fig. 1B). These indicate that upon faster depletion of PI(4,5)P₂ in oocytes coexpressing K555Anap with WT Ci-VSPs than in those expressing K555Anap Ci-VSP alone (SI Appendix, Fig. S1D), Anap fluorescence showed more rapid decay time course, consistent with the idea that the fluorescence kinetics of K555Anap correlates with the time course of the availability of PI(4,5)P₂. Taken together, changes in the time course of Anap fluorescence serve as a suitable indicator for PI(4,5)P₂-dependent changes of the local environment occurring at a specific site within Ci-VSP.

Structural Rearrangements of Basic Residues within the VSD-PD Linker Are Sensitive to PI(4,5)P₂.

K555Anap fluorescence results inspired us to expect that interaction between the VSD-PD linker of VSP and PI(4,5)P₂ could be explored by a similar experiment. For this purpose, we incorporated Anap on K252, R253, or R254 which are known to be critical for VSD-CCR coupling (16, 17) (Fig. 2A). However, there was no difference in the fluorescence time courses between the long and the short interval protocols (SI Appendix, Fig. S3 A and B). Because these residues are also critical for phosphatase activity (17), we examined voltage-dependent phosphatase activities of the Anap mutants by using a PI(4,5)P₂-sensitive inwardly rectifying K⁺ (K_ir2.1) channel as a reporter of VSP-induced changes of PI(4,5)P₂ levels in Xenopus oocytes (SI Appendix, Fig. S3C): The rate constant determined by fitting the normalized K_ir current decay upon repetitive depolarization with a single exponential function was taken as a reporter of the phosphatase activity (3, 26). Although cells coexpressing WT Ci-VSP showed a rapid current decrease upon repeated depolarization to 50 mV, all three Anap mutants induced little current decrease even in response to depolarization to 160 mV (SI Appendix, Fig. S3 C and D), indicating weaker phosphatase activity compared to the WT (SI Appendix, Fig. S3E). All three Anap mutants exhibited robust cell surface expression as evidenced by voltage-evoked sensing currents (1) (SI Appendix, Fig. S3F) with a small leftward shift of the moving charges of the off sensing currents (Q_OFFs) against membrane potential (Q_OFF-V relationship) compared with the WT (SI Appendix, Fig. S3G).

Then WT Ci-VSP was coexpressed with each of the Anap mutants to deplete PI(4,5)P₂. Neither K252Anap nor R253Anap exhibited any difference in the fluorescence time courses between the long and the short interval protocols (SI Appendix, Fig. S3 H and I). On the other hand, R254Anap showed a difference in the fluorescence time courses at depolarizing pulse between the long and the short interval protocols in the same oocyte (SI Appendix, Fig. S3 H and I), indicating PI(4,5)P₂-dependent change of R254Anap fluorescence time course. F_peak − F_ss was significantly increased by coexpression with WT Ci-VSP (SI Appendix, Fig. S3J). These results suggest that PI(4,5)P₂ is located in close proximity to R254 consistent with the previous reports (16–18).

Five basic residues (K241, R245, R246, R256, and K257) within the VSD-PD linker (SI Appendix, Fig. S4A) were similarly studied. We found that K257Anap exhibited a hump-shaped change of the fluorescence time course in response to the long interval protocol (SI Appendix, Fig. S4B). This change was not observed by the short interval (SI Appendix, Fig. S4B), suggesting that the fluorescence time course of K257Anap is also sensitive to availability of PI(4,5)P₂. Four other residues did not show any change in Anap fluorescence time course upon depolarization of the long interval protocol (SI Appendix, Fig. S4B).

Structural Rearrangements of Noncharged Residues, A242 and Y255, within the VSD-PD Linker Are Also Sensitive to PI(4,5)P₂.

A previous study reported that cysteine substitution of A242, R245, N251, Y255, and R256 within the VSD-PD linker (SI Appendix, Fig. S4A) abrogated phosphatase activity (27), suggesting that noncharged residues such as A242, N251, and Y255 also play a crucial role in VSD-CCR coupling. To test whether these are involved in interaction with PI(4,5)P₂, fluorescence changes of Anap incorporated into these three noncharged residues observed in the long and the short interval protocols were compared.

Oocytes expressing Y255Anap Ci-VSP (Fig. 2A) exhibited a drastic change of the fluorescence time course (F_peak) in response to membrane depolarization of the long interval protocol: a fast and large phasic decrease which peaked before going into an increase to a steady state (F_ss) at the end of depolarization (Fig. 2B, red arrow). Characteristically, there were two relatively small fluorescence components different from F_peak: a fast transient increase right after depolarization (F_initial) and a gradual decrease to the original level right after repolarization to −60 mV (F_repo) (Fig. 2B, cyan and magenta arrow heads, respectively). The fluorescence-voltage (F-V) relationship of F_ss showed biphasic voltage dependence: a small increase at 20 to 80 mV and an apparent decrease at higher membrane potentials (SI Appendix, Fig. S5A). In the short interval protocol, F_peak − F_ss, F_initial, and F_repo significantly decreased in the same oocyte (Fig. 2B, black trace, and SI Appendix, Fig. S5B), suggesting that Y255Anap signal consists of PI(4,5)P₂-dependent fluorescence components. When WT Ci-VSP was coexpressed with Y255Anap Ci-VSP to enhance phosphatase activity in the oocyte (SI Appendix, Fig. S3E), both the time to peak of the fluorescence decrease (F_peak) from the peak of F_initial (t_peak = 50.9 ± 5.0 ms and 94.3 ± 20.3 ms for Y255Anap with and without WT, respectively) and τ of the fluorescence increase from F_peak determined by fitting it with a single exponential function (41.8 ± 4.3 ms versus 113.5 ± 61.9 ms) were significantly shortened (Fig. 2 C and D and SI Appendix, Fig. S5C). These results indicate that the change of Y255Anap fluorescence time course (F_peak) is influenced by depletion of PI(4,5)P₂.

We also recorded Anap fluorescence from oocytes expressing the Y255Anap mutant which contains the enzymatically inactive mutation, C363S (C363S/Y255Anap Ci-VSP). C363S/Y255Anap exhibited only a fluorescence decrease that was followed by F_initial and reached a steady state at the end of depolarization in the long interval protocol (SI Appendix, Fig. S6A, red trace). As seen in the F-V curve (SI Appendix, Fig. S6B), C363S/Y255Anap Ci-VSP showed a similar biphasic fluorescence change in F_ss to that of Y255Anap with no change in the Q_OFF-V relationship (SI Appendix, Fig. S6C). On the other hand, F_repo was not observed in C363S/Y255Anap Ci-VSP (SI Appendix, Fig. S6A), indicating that structural rearrangements of Y255 are different between enzymatically active and inactive forms of Ci-VSP. In oocytes coexpressing C363S/Y255Anap with WT Ci-VSPs, however, depolarizing pulses failed to induce a change in both F_peak − F_ss and F_initial (SI Appendix, Fig. S6D). Since the C363S mutation traps the substrate in the active site and prevents substrate turnover, these suggest that substrate turnover couples with PI(4,5)P₂-dependent structural rearrangements of Y255.

Furthermore, we incorporated Anap on A242 and N251. A242Anap did not show a difference in the fluorescence time courses between the long and the short interval protocols (SI Appendix, Fig. S4C). However, at a bandwidth of 420 to 460 nm, A242Anap showed a difference in the time courses between the two protocols (SI Appendix, Fig. S4C). In N251Anap, fluorescence changes did not show any difference in the fluorescence time courses between the two protocols at both the 460- to 510-nm and 420- to 460-nm bandwidths (SI Appendix, Fig. S4C). Collectively, Anap incorporated at noncharged residues, A242 and Y255, within the C-terminal part of the VSD-PD linker reports PI(4,5)P₂-dependent local structural rearrangements.

Anap Incorporated into the VSD-PD Linker Senses Nonsubstrate PI(4,5)P₂ Outside the Active Site.

As K555 is located near the active site pocket (SI Appendix, Fig. S7A), it is likely that PI(4,5)P₂-dependent time course of K555Anap fluorescence changes mainly reflects interaction with the substrate PI(4,5)P₂ at the active site. On the other hand, at A242, R254, Y255, and K257, fluorescence changes of Anap may reflect interaction with a PI(4,5)P₂ molecule associated with the VSD-PD linker which is not identical to the substrate one at the active site. To test this hypothesis, we compared fluorescence changes between Y255Anap and K555Anap in more detail.

First, we compared the F-V relationships of F_peak between Y255Anap and K555Anap. The F_peak-V curve for Y255Anap qualitatively differed from that for K555Anap: K555Anap peaked at 140 mV and decreased F_peak at higher membrane potentials, whereas Y255Anap only gradually increased F_peak with increased membrane potential up to 180 mV (Fig. 3A). The V_1/2 of the Q_OFF-V curve, which gives half of the maximum change of Q_OFF, was slightly shifted rightward by about 14.7 mV in Y255Anap (SI Appendix, Fig. S7B) which might not account for a pronounced shift of the F_peak-V curve of Y255Anap.

Fig. 3.

PI(4,5)P₂-dependent structural rearrangements of R254, Y255, and K257 in a mutant with low active site occupancy for PI(4,5)P₂. (A) F_peak-V curves for K555Anap and Y255Anap. The fluorescence value at −60 mV (F) was subtracted from F_peak (ΔF) of each membrane potential, and then ΔF was normalized to F (ΔF/F). ΔF/F of each membrane potential was again normalized to the maximal ΔF/F in the same recording. Data are shown as mean ± SD; n = 5 and 7 for K555Anap and Y255Anap, respectively. (B) Representative fluorescence traces of G365A/K555Anap from the same oocyte expressed alone (Left) and coexpressed with WT Ci-VSP (Right) in response to step to 160 mV recorded using the long (red and blue) and the short (black) interval protocols were superimposed. Traces are averaged from five runs. Inset: Schematic representation of G365A Ci-VSP showing low PI(4,5)P₂ occupancy at the active site. (C and D) Left: Representative fluorescence traces of G365A/Y255Anap from the same oocyte expressing G365A/Y255Anap Ci-VSP alone (C) and coexpressing with WT Ci-VSP (D) in response to step to 160 mV recorded using the long (red and blue) and the short (black) interval protocols were superimposed. The blue arrow in D indicates changes in the time course recorded with long interval. Traces are averaged from five runs. Right: Changes in the amplitudes of F_peak − F_ss recorded with long (L) and short (S) intervals. F_peak − F_ss values were normalized to F_ss values. Bars are means ± SD; n = 5 and 4 for G365A/Y255Anap without and with WT Ci-VSP, respectively; two-tailed paired t test. (E) Comparison of G365A/Y255Anap fluorescence traces upon depolarization to 160 mV of the long interval protocol in the absence (red, from C) and the presence (blue, from D) of WT Ci-VSP (Left) and F_peak − F_ss (Right). Bars are means ± SD; n = 5 and 4 for G365A/Y255Anap without and with WT Ci-VSP; *P < 0.05, two-tailed unpaired t test. (F and G) Left: Representative fluorescence traces of G365A/R254Anap (F) and G365A/K257Anap (G) from the same oocyte coexpressing with WT Ci-VSP in response to step to 160 mV recorded using the long (blue) and the short (black) interval protocols were superimposed. Blue arrows indicate changes in the time course recorded with long interval. Traces for G365A/R254Anap and G365A/K257Anap are averaged from three and five runs, respectively. Right: Changes in the amplitudes of F_peak − F_ss recorded with long (L) and short (S) intervals. F_peak − F_ss values were normalized to F_ss values. Bars are means ± SD; n = 4 and 3 for G365A/R254Anap and G365A/K257Anap, respectively; *P < 0.05, two-tailed paired t test. The fluorescence was detected using a 460- to 510-nm band-pass emission filter.

Next, we measured fluorescence changes of Anap at residues K555 and Y255 with the background of G365A mutation. G365 is a part of the active site pocket of Ci-VSP (25, 28) and corresponds to A126 in human PTEN, where various amino acid mutations have been found in human cancer patients (COSMIC database) (29). In the previous studies of Ci-VSP, G365A mutation strongly decreased 5-phosphatase activity toward PI(4,5)P₂ (5, 25, 30). Therefore, we hypothesized that the occupancy of PI(4,5)P₂ at the active site is low or lost in G365A Ci-VSP. Oocytes expressing neither G365A/K555Anap Ci-VSP alone nor G365A/K555Anap and WT Ci-VSPs exhibited any difference in the fluorescence time courses during membrane depolarization between the long and the short interval protocols (Fig. 3B and SI Appendix, Fig. S8A). Comparisons between G365A/K555Anap and K555Anap Ci-VSPs showed a similar Q_OFF-V relationship and characteristic biphasic fluorescence profile of F_ss upon membrane depolarization: a small decrease at 0 to 100 mV and a large increase at higher membrane potentials (3, 21, 22) (SI Appendix, Fig. S8 B and C). These support the idea that local structural rearrangements of K555 are affected by the substrate PI(4,5)P₂ at the active site.

G365A mutation also diminished the difference in the time course of F_peak of Y255Anap between the long and the short interval protocols, consistent with the lack of PI(4,5)P₂ 5-phosphatase activity in G365A Ci-VSP (Fig. 3C, red trace). Unlike Y255Anap without G365A, the clear downward peak (F_peak), which represents sensitivity to PI(4,5)P₂ depletion as mentioned above (Fig. 2B, red trace), was not observed. F_initial and F_repo in response to membrane depolarization to 160 mV and repolarization to −60 mV, respectively (Fig. 3C), were similar to those observed in Y255Anap without G365A, suggesting that structural rearrangements of Y255 in G365A Ci-VSP underlying these fluorescence changes are similar to those in WT Ci-VSP. In contrast with G365A/K555Anap, the Q_OFF-V curve was shifted rightward compared with Y255Anap Ci-VSP without G365A, and the negative second component of the F_ss-V curve was also shifted toward more positive voltages with an expansion of the first positive component to a wider range of voltages (SI Appendix, Fig. S8 D and E).

WT Ci-VSP was then coexpressed with G365A/Y255Anap Ci-VSP to deplete PI(4,5)P₂. We observed a change in the fluorescence time course during membrane depolarization in the long interval protocol: a small increase that was followed by a decrease from F_initial (Fig. 3D, blue trace). When the same oocyte was recorded with the short interval protocol, F_peak − F_ss, F_initial, and F_repo were decreased (Fig. 3D and SI Appendix, Fig. S8F), suggesting PI(4,5)P₂-dependent changes in the time course of Y255Anap in G365A Ci-VSP. A significant increase in F_peak − F_ss accompanied with a significant decrease in F_initial and F_ss in the long interval protocol was induced by coexpression with WT Ci-VSP (Fig. 3E and SI Appendix, Fig. S8G). Coexpression of the type I phosphatidylinositol-4-phosphate 5-kinase (PIP5K), which converts phosphatidylinositol 4-phosphate (PI(4)P) to PI(4,5)P₂, diminished the decrease in Y255Anap fluorescence from F_initial and F_repo (SI Appendix, Fig. S8H, green trace), indicating the involvement of PI(4,5)P₂ in the Y255Anap fluorescence changes upon Ci-VSP activation. Collectively, these results support the idea that Anap on Y255 can sense nonsubstrate PI(4,5)P₂ outside the active site unlike that on K555.

To see whether Anap incorporated at the other residues within the VSD-PD linker also senses nonsubstrate PI(4,5)P₂, we examined the effects of G365A mutation on PI(4,5)P₂-sensitive changes in the fluorescence of A242Anap, R254Anap, and K257Anap. Anap fluorescence at all the three residues in the G365A mutant did not exhibit any difference in the fluorescence time courses between the long and the short interval protocols (SI Appendix, Fig. S9 A and B). When the WT Ci-VSP was coexpressed with each of the mutants, both G365A/R254Anap and G365A/K257Anap exhibited a difference in the time courses of Anap fluorescence between the long and the short interval protocols (Fig. 3 F and G, blue trace). Oocytes coexpressing G365A/A242Anap with WT Ci-VSPs, however, did not show any difference in Anap fluorescence time courses between the long and the short interval protocols (SI Appendix, Fig. S9C). These results indicate that not only Y255 but multiple residues within the VSD-PD linker centering around Y255 sense nonsubstrate PI(4,5)P₂ outside the active site.

Anap Incorporated at the Tyrosine Residue within the VSD-PD Linker in PI(3,4,5)P₃ 3-Phosphatase, Ci-VSPTEN Chimera, Reports Structural Rearrangements Sensitive to Nonsubstrate PI(4,5)P₂.

We also analyzed a chimeric protein which is composed of the VSD and the N-terminal part of the VSD-PD linker of Ci-VSP followed by the mouse PTEN, and exhibits voltage-dependent PTEN-like 3-phosphatase activity toward PI(3,4,5)P₃ (24, 31) (Fig. 4A). A human PTEN construct was used in this study [for this subtle change from the construct generated in the original work (24), our construct was renamed as Ci-VSPTEN16h]. While N-terminus of PTEN does not have lysine at a site equivalent to K257 in Ci-VSP, other residues at which Anap shows PI(4,5)P₂-dependent change of the fluorescence time course in Ci-VSP (A3, R15, and Y16 corresponding to A242, R254, and Y255 in Ci-VSP, respectively) are conserved in Ci-VSPTEN16h (Fig. 4A). The tyrosine residue of Ci-VSPTEN16h corresponding to Y255 of Ci-VSP (equivalent to Y16 in PTEN) was replaced by Anap (Y255Anap Ci-VSPTEN16h) and its Anap fluorescence was measured under conditions of normal or depletion states of PI(4,5)P₂. Y255Anap Ci-VSPTEN16h showed a fluorescence decline during 500-ms depolarizing step with the long interval protocol, which lacked F_initial, F_peak, and F_repo compared to Y255Anap in Ci-VSP (Fig. 4B, red trace). In oocytes coexpressing with WT Ci-VSP, depolarization elicited a change in the fluorescence time course with the long interval protocol, which was suppressed with the short interval protocol in the same oocyte (Fig. 4C, blue trace). This indicates PI(4,5)P₂ dependence of the fluorescence time course of Anap at the site of Ci-VSPTEN16h equivalent to Y255 of Ci-VSP. F_peak − F_ss was significantly increased by coexpression with WT Ci-VSP (Fig. 4D). These support again our idea that nonsubstrate PI(4,5)P₂ interacts with the VSD-PD linker of Ci-VSP.

Fig. 4.

Nonsubstrate PI(4,5)P₂-dependent structural rearrangements of Y255 in Ci-VSPTEN. (A) Upper: Schematic of the domain organization of Ci-VSP, hPTEN, and Ci-VSPTEN16h chimera. C-tail, disordered C-terminal tail. Lower: Sequence alignment of the VSD-PD linker of Ci-VSP (magenta), the N-terminus of hPTEN (cyan), and the linker of Ci-VSPTEN16h. Numbers refer to the amino acid positions in Ci-VSP, hPTEN, and Ci-VSPTEN16h, from Top to Bottom respectively. A242, R254, and Y255 in Ci-VSP (black dots) are conserved in Ci-VSPTEN16h (black arrows) but K257 (black dot) is lacked in Ci-VSPTEN16h. Inset: Schematic representation of Ci-VSPTEN16h in which PI(3,4,5)P₃ instead of PI(4,5)P₂ is bound to the active site. (B and C) Left: Representative fluorescence traces of Y255Anap in Ci-VSPTEN16h from the same oocyte expressing Ci-VSPTEN16h alone (B) and coexpressing with WT Ci-VSP (C) in response to step to 160 mV recorded using the long (red and blue) and the short (black) interval protocols were superimposed. The blue arrow in C indicates a change in the time course recorded with long interval. The fluorescence was detected using a 460- to 510-nm band-pass emission filter. Traces are averaged from five runs. Right: Changes in the amplitudes of F_peak − F_ss recorded with long (L) and short (S) intervals. F_peak − F_ss values were normalized to F_ss values. Bars are means ± SD; n = 7 and 5 for Y255Anap in Ci-VSPTEN16h without and with WT Ci-VSP; two-tailed paired t test. (D) Comparison of Y255Anap fluorescence traces in Ci-VSPTEN16h upon depolarization to 160 mV of the long interval protocol in the absence (red, from B) and the presence (blue, from C) of WT Ci-VSP (Left) and F_peak − F_ss (Right). Bars are means ± SD; n = 7 and 5 for Y255Anap in Ci-VSPTEN16h without and with WT Ci-VSP; *P < 0.05, two-tailed unpaired t test.

Phenolic Ring Structure Plays a Critical Role of Residue Y255 in VSD-CCR Coupling.

One may wonder whether the fluorescence change of Anap on Y255 does not reflect direct PI(4,5)P₂ interaction with Y255Anap, but rather indirectly reflects global structural rearrangements of the VSD-PD linker which could be mediated by PI(4,5)P₂ interaction with a site remote from Y255. To investigate the possibility that PI(4,5)P₂ interaction with a positively charged residue adjacent to Y255 (R256; SI Appendix, Fig. S10A) affects PI(4,5)P₂-sensitive changes in the fluorescence of Y255Anap, we neutralized this residue in Y255Anap Ci-VSP. R256Q/Y255Anap Ci-VSP showed a similar change in the fluorescence time course upon depolarization of the long interval protocol to that of Y255Anap Ci-VSP without R256Q mutation (SI Appendix, Fig. S10B, red trace), suggesting that structural rearrangements of Y255 are insensitive to PI(4,5)P₂ which interacts with R256.

The effect of mutation in R254, one position N-terminally from Y255, on fluorescence change of Anap on Y255 was similarly tested. Charge-neutralizing mutations in R254 (glutamine, alanine, and glycine) abolished F_initial, F_peak, and F_repo of Y255Anap fluorescence which did not show any difference in the time courses between the long and the short interval protocols with a shift of the Q_OFF-V relationship (SI Appendix, Fig. S10 C and D). This could be probably due to disruption of structure of the VSD-PD linker, which is critical for transmission of structural change of the VSD to the CCR, as suggested by our previous finding that voltage-dependent structural rearrangements of the CCR reported by Anap are known to be abolished by R253A/R254A double mutation (21).

To gain further insights into the contribution of Y255 in VSD-CCR coupling, we substituted Y255 for various amino acids. Phosphatase activities of the mutants at 50 mV were compared with that of WT by coexpressing with a PI(4,5)P₂-sensitive K_ir channel, GIRK2d, as established previously (26). Y255F as well as Y255W showed similar strength of phosphatase activity to WT (Fig. 5A and SI Appendix, Fig. S12A), consistent with the result of an early study which reported that phenylalanine substitution of Y255 maintained phosphatase activity (27). By contrast, other mutants attenuated phosphatase activity even though cell surface expression of them was robust and a rightward shift in the Q_OFF-V relationship compared to WT was insufficient to explain it (Fig. 5A and SI Appendix, Fig. S12 A–C). Interestingly, substitutions of Y255 with basic amino acids (Y255K and Y255R) drastically decreased phosphatase activity (Fig. 5A), while neighboring residues on Y255 are enriched with positively charged amino acids (Fig. 2A). These results suggest that the phenolic ring structure at residue 255 is important for phosphatase activity probably through direct interaction with PI(4,5)P₂.

Fig. 5.

Effects of mutations at Y255 on phosphatase activity and voltage-driven conformational changes in S4. (A) Summary of phosphatase activity at 50 mV for WT and eleven Y255 mutants. K, R, and A mutants did not exhibit any phosphatase activities. Bars are means ± SD; n ≥ 4. (B) Representative TMRM fluorescence traces of Q208C* (cartoon Inset), Y255F/Q208C*, and Y255R/Q208C* in response to depolarizing pulses (Top) to potentials indicated by colors. (C) F-V curves for Q208C* (Left), Y255F/Q208C* (Center), and Y255R/Q208C* (Right). The fluorescence values measured at arrows in B (average values between 200 ms and 210 ms) were normalized to that at 0 mV. Data are shown as mean ± SD; n = 6, 5, and 7 for Q208C*, Y255F/Q208C*, and Y255R/Q208C*, respectively.

Effects of amino acid change at Y255 on coupling states between the VSD and the CCR were then examined by fluorescence changes in a thiol reactive fluorescent dye, tetramethylrhodamine-6-maleimide (TMRM), attached to a cysteine residue substituted at 208 within the S3-S4 extracellular loop (Q208C). Q208C-TMRM (Q208C*) in Ci-VSP has been reported to show a triphasic fluorescence change voltage-dependently: an initial small increase at the holding potential of −60 mV to 30 mV, a decrease at 40 to 100 mV, and an increase at voltages higher than 100 mV (3, 17, 32, 33) (Fig. 5 B and C). The third increasing fluorescence component reflects voltage-induced structural rearrangements of S4 highly sensitive to VSD-CCR coupling and was attenuated by neutralization mutations K252Q, R253Q, and R254Q of the VSD-PD linker (17). Either mutation of Y255F, which maintains phosphatase activity and might keep interaction with PI(4,5)P₂, or Y255R, which inhibits phosphatase activity and might prevent interaction with PI(4,5)P₂, was individually introduced into Q208C Ci-VSP. Y255F mutation did not affect all the three fluorescence components (Fig. 5 B and C). Y255R mutation did not affect the first fluorescence component but increased the amplitude of the fluorescence change in the second decreasing component and suppressed the third component (Fig. 5 B and C), consistent with results of voltage-induced enzyme activities (Fig. 5A). Taken together, these findings indicate that PI(4,5)P₂ interaction with the VSD-PD linker plays a critical role in VSD-CCR coupling.

Direct PI(4,5)P₂ Interaction with the VSD-PD Linker Suggested by Molecular Dynamics Simulations.

To further assess interaction between the VSD-PD linker of Ci-VSP and nonsubstrate PI(4,5)P₂, we conducted all-atom MD simulations using the predicted activated state structure of Ci-VSP (3). VSP was embedded in a 1-palmitoyl-2-oleoyl-sn-glycero-3-phosphocholine (POPC) lipid bilayer with different concentrations of PI(4,5)P₂ in the cytoplasmic leaflet. At a concentration of 15% PI(4,5)P₂ with their randomized starting positions, a 495-ns MD simulation showed that PI(4,5)P₂ located near S4 was drawn into the cytoplasm and approached the C-terminal part of the VSD-PD linker (Fig. 6 A and B and SI Appendix, Fig. S13A). The side chains of Y255 and K257, at which Anap exhibited PI(4,5)P₂-dependent change in the fluorescence time course (Fig. 2B and SI Appendix, Fig. S4B), were in close proximity to PI(4,5)P₂ with distances of 6.2 Å between C-4 of the phenolic ring and P5-phosphate and 4.4 Å between C_ε and P5-phosphate, respectively (t2 = 188.5 ns; Fig. 6 A and B). During the second half of this simulation, two PI(4,5)P₂ molecules simultaneously approached the C-terminal part of the VSD-PD linker: Residues 254 to 257 were localized in the proximity of any of two PI(4,5)P₂ molecules (SI Appendix, Fig. S13 A–C). There was no PI(4,5)P₂ molecule at the active site which moved to or from space close to the VSD-PD linker. In the second replica, while the side chain of Y255 did not face toward PI(4,5)P₂ (the minimal distance of 8.2 Å between C-4 of the phenolic ring and P4-phosphate at 177.0 ns), we observed that PI(4,5)P₂ formed stable interactions with R256 and K257 during the first half of the simulation (Fig. 6 C and D and SI Appendix, Fig. S13D). The PI(4,5)P₂ molecule also interacted with the N-terminal part of the VSD-PD linker during the second half of the simulation with distances of 6.1 Å between C_β of A242 and P5-phosphate and 3.8 Å between C_ζ of R246 and P5-phosphate at 320.5 ns (Fig. 6 C and D and SI Appendix, Fig. S13D). These replicate simulations suggest that the VSD-PD linker is capable of interacting with nonsubstrate PI(4,5)P₂, supporting our results of VCF with Anap.

Fig. 6.

PI(4,5)P₂ interaction with multiple sites within the VSD-PD linker of Ci-VSP in MD simulations. (A) Plots for distances between P5-phosphate of PI(4,5)P₂ and the farthest side chain carbon atom from C_α atom of residues of interest within the VSD-PD linker against simulation time (see also SI Appendix, Fig. S13). (B) Snapshots from the indicated time points in A. PI(4,5)P₂ of interest in the simulation and C_α atom of C363 are shown as a stick (green) and a sphere (magenta), respectively. S4 and the VSD-PD linker are colored blue and green, respectively. Side chains are shown as sticks for residues shown in A. (C) Similar distance plots from the second replica. The Lower panel is an enlarged time scale (from 0 to 20 ns) of the Upper panel. (D) Snapshots from the indicated time points in C.

Simulations were then performed under lower concentrations of PI(4,5)P₂. A 250-ns short simulation where Ci-VSP was placed in a POPC membrane with a 10% concentration of PI(4,5)P₂ in the cytoplasmic leaflet revealed stable interactions between the N-terminal part of the VSD-PD linker and PI(4,5)P₂ for more than 200 ns (SI Appendix, Fig. S13 E and F). Stable PI(4,5)P₂ interactions with the VSD-PD linker were predominantly coordinated by A242, R245, R246, R254, R256, and K257 in two simulations with a concentration of 5% PI(4,5)P₂ (SI Appendix, Fig. S13 G–J). On the other hand, one out of three simulations with a 5% concentration of PI(4,5)P₂ did not show any interaction between the VSD-PD linker and PI(4,5)P₂, in which no PI(4,5)P₂ molecule was localized in proximity to S4 throughout the simulation. This indicates that simulations are occasionally affected by a starting position of PI(4,5)P₂ on the cytoplasmic leaflet due to a limitation of our simulations in which PI(4,5)P₂ molecules were randomly placed at the beginning of each simulation.

Overall, our simulations reproducibly showed interactions between residues within the VSD-PD linker of Ci-VSP and nonsubstrate PI(4,5)P₂ which is independent of the concentration of PI(4,5)P₂ in the cytoplasmic leaflet. This well supports our findings in VCF experiments that R254, Y255, and K257 within the VSD-PD linker corresponding to the N-terminus of PTEN have interaction with nonsubstrate PI(4,5)P₂. These simulations also suggest that basic residues [K241, R245, R246, and R256 at which Anap had no difference in the fluorescence time courses between the long and the short interval protocols (SI Appendix, Fig. S4B)] support interaction of the VSD-PD linker with PI(4,5)P₂.

Discussion

In this study, using an environmentally sensitive, fluorescent unnatural amino acid, Anap, we comprehensively examined residues within the VSD-PD linker of Ci-VSP where PI(4,5)P₂ has been proposed to interact for coupling between voltage sensing and catalytic activity (16, 17). Results of the PI(4,5)P₂-dependent change in the fluorescence time course of Anap on A242, R254, Y255, and K257 upon membrane depolarization in the active form enzyme suggest that the VSD-PD linker interacts with PI(4,5)P₂. R254Anap, Y255Anap, and K257Anap showed the PI(4,5)P₂-dependent change in the fluorescence time course even in the background of G365A mutation where the active site occupancy for the substrate PI(4,5)P₂ is low or lost, whereas K555Anap in proximity to the active site did not, supporting that the linker interacts with nonsubstrate regulatory PI(4,5)P₂ outside the active site. This view is also supported by the facts that a shift of the F_ss-V curve of Y255Anap and K555Anap upon introduction of G365A mutation was to the opposite direction along the voltage axis between the two Anap-incorporated sites (SI Appendix, Fig. S8 B and D). The PI(4,5)P₂-dependent change in the time course of Y255Anap was also found in PI(3,4,5)P₃ 3-phosphatase Ci-VSPTEN chimera which does not dephosphorylate PI(4,5)P₂. MD simulations showed PI(4,5)P₂ docked near residues within the VSD-PD linker including Y255. Taken together, our findings indicate that the VSD-PD linker of VSP corresponding to the PI(4,5)P₂-binding region at the N-terminus of PTEN interacts with nonsubstrate PI(4,5)P₂ to regulate VSD-CCR coupling (Fig. 7A).

Fig. 7.

PI(4,5)P₂ regulation of phosphatase activities of VSP and PTEN. (A) Schematic model for VSD-CCR coupling mediated by nonsubstrate, regulatory PI(4,5)P₂ in Ci-VSP. A bold gray horizontal line indicates the cytoplasmic side of the plasma membrane. (B) Phosphatase activity of VSP is suppressed through negative feedback loop due to PI(4,5)P₂ depletion, whereas that of PTEN is enhanced through positive feedback loop resulted from PI(4,5)P₂ production.

Physicochemical Mechanism Underlying Interaction between PI(4,5)P₂ and the VSD-PD Linker.

Given that anionic phospholipids, such as PI(4,5)P₂, form a salt bridge with basic residue in a protein, R254 and K257 within the VSD-PD linker of Ci-VSP may tightly interact with PI(4,5)P₂. In fact, our MD simulations showed distances of 4.1 Å between C_ζ of R254 and P5-phosphate of PI(4,5)P₂ at 370.5 ns (SI Appendix, Fig. S13B), and 4.0 Å and 3.9 Å between C_ε of K257 and P5-phosphate of PI(4,5)P₂ at 142.5 ns and 12.0 ns (Fig. 6 A and C), respectively, consistent with the definition of salt bridge formation between N-O atom pairs (≤4 Å) (34).

Less than 4 Å distance between R256 and PI(4,5)P₂ in the simulations (Fig. 6C and SI Appendix, Fig. S13 B, G, and I) suggests that PI(4,5)P₂ interaction with R256, a basic residue on the C-terminal side of Y255, mediates the Y255Anap fluorescence changes since Anap could report changes in the local environment secondarily induced by structural rearrangements of a residue remote from the Anap incorporation site (35). Nevertheless, we assume that Y255 could directly interact with PI(4,5)P₂. Distance near 6 Å (36, 37) between Y255 and PI(4,5)P₂ in the simulation shown in Fig. 6A presented a potential for cation- or anion-π interactions. Maintenance of phosphatase activity in Y255Anap Ci-VSP indicates that Anap comprising the benzene ring also establishes interaction with PI(4,5)P₂, supporting the idea that the benzene ring structure at residue 255 is critical to interact with PI(4,5)P₂. This is consistent with phosphatase activities and VSD-CCR coupling assessed by Q208C* fluorescence changes of amino acid mutants at Y255 Ci-VSP, where aromatic substitutions maintained phosphatase activity and coupling, whereas other mutants decreased phosphatase activity and Y255R mutant lacked the coupling-related fluorescence component (Fig. 5). In fact, R256Q mutation did not prohibit the characteristic fluorescence change of Y255Anap time course (SI Appendix, Fig. S10B), supporting that interaction of PI(4,5)P₂ with Y255 is independent of R256. We also tried to verify the possible effect of R254 on Y255Anap fluorescence changes. Charge-neutralizing mutations in R254 had a strong impact on the fluorescence signal from Y255Anap with a shift of the Q_OFF-V relationship (SI Appendix, Fig. S10 C and D). Previous studies showed that R254 is one of the critical residues required for transmission of structural change of the VSD to the CCR (17). The voltage-dependent motion of the CCR has been shown to be completely silenced by R253A/R254A mutation (21). We assume that R254 possibly in concert with R253 maintains the linker structure which is required for PI(4,5)P₂ interacting with Y255. However, further studies are required to clarify the detailed dynamics of PI(4,5)P₂ on the VSD-PD linker and profile of interaction between PI(4,5)P₂ and residues within the linker.

Physiological Role of Residence of PI(4,5)P₂ on the VSD-PD Linker.

Conserved from eukaryotes to humans, consecutive arginine-tyrosine motif is situated at the C-terminal half of the VSD-PD linker of VSP and at the N-terminus of PTEN (corresponding to R254 and Y255 in Ci-VSP, and R15 and Y16 in human PTEN, respectively; SI Appendix, Fig. S1). This suggests that interaction of nonsubstrate PI(4,5)P₂ with the linker of VSP and the N-terminus of PTEN plays a pivotal role in regulation of phosphatase activity. Upon membrane depolarization, VSP dephosphorylates mainly PI(4,5)P₂ and consequently the total number of PI(4,5)P₂ on the plasma membrane is decreased. As our results showed that PI(4,5)P₂ interaction with the linker is necessary to VSD-CCR coupling (Fig. 5), phosphatase activity might be suppressed through a negative feedback loop (Fig. 7B). Since PI(4,5)P₂ is concentrated at the plasma membrane (11), PI(4,5)P₂-mediated VSD-CCR coupling might have another mechanism by which VSP is only activated on the plasma membrane and silenced in endomembranes, which contain little PI(4,5)P₂ (11). In mouse sperm, VSP could be weakly activated at the resting membrane potential but the flagellum showed a heterogeneous PI(4,5)P₂ distribution (6). Along such biological context, the negative feedback loop could be important to maintain phosphatase activity of VSP at a physiologically appropriate level and control PI(4,5)P₂ distribution in the sperm.

On the other hand, PTEN dephosphorylates PI(3,4,5)P₃ which leads to production of PI(4,5)P₂ and PI(4,5)P₂ positively modulates phosphatase activity of PTEN (13–15) (Fig. 7B). A study using social amoebae Dictyostelium discoideum showed that acceleration of PI(3,4,5)P₃ dephosphorylation of PTEN at only the side and the posterior parts of the cell through PI(4,5)P₂-mediated positive feedback restricted distribution of PI(3,4,5)P₃, which is a signal molecule for an extension of pseudopod, to the anterior, stabilizing the anterior–posterior polarization and efficient migration of the cell (38). Within the putative PI(4,5)P₂-binding region at the N-terminus, K13 and R14 as well as R15 and Y16 (corresponding to K252 to Y255 of Ci-VSP, respectively) are all hot spots of genetic mutation in human cancer patients (COSMIC database), indicating physiological importance of PI(4,5)P₂ binding to the N-terminus of PTEN. It is intriguing that regulation of phosphatase activity through regulatory PI(4,5)P₂ interaction has the opposite physiological significance in these two structurally similar phosphoinositide phosphatases, VSP and PTEN (Fig. 7B).

Relevance to PI(4,5)P₂ Regulation of Voltage-Gated Ion Channels.

In the case of diverse VGICs, PI(4,5)P₂ reportedly regulates their functions positively or negatively (10–12). Detailed structural basis for PI(4,5)P₂ interaction with VGICs has been identified by cryo-EM or predicted by MD simulations (39–44). In voltage-gated potassium channel K_v7.1 (KCNQ1), PI(4,5)P₂ interacted with positively charged amino acids within the S4-S5 linker, which connects the VSD with the pore-gate domain (PGD), and the interaction was suggested to regulate pore opening through coupling between the VSD and the PGD (39–41). Structural rearrangements of the VSD induces PI(4,5)P₂ interaction with the linker in a state-dependent manner (41, 45). Lack of PI(4,5)P₂-dependent Anap fluorescence change at Y255 in the enzymatically inactive C363S Ci-VSP mutant is consistent with a state-dependent PI(4,5)P₂ interaction with the VSD-PD linker of VSP. State-dependent interaction of PI(4,5)P₂ with a linker connecting the VSD with a downstream effector domain could be a common mechanism underlying transduction of electrical signal detected by the VSD among the voltage-dependent membrane protein superfamily.

Materials and Methods

Molecular Biology.

The WT Ci-VSP was in pSD64TF vector (1). Amino acid mutations were generated by using the PrimeSTAR Mutagenesis Basal Kit (Takara Bio Inc., Japan) and confirmed by DNA sequencing. The stop codon of WT Ci-VSP [amber nonsense codon (TAG)] was replaced with the ochre nonsense codon (TAA). Mouse K_ir3.2d (GIRK2d) (46), and bovine G-protein β and γ (Gβ and Gγ) (47) were kind gifts from Yoshihisa Kurachi (The University of Osaka, Osaka, Japan) and Toshihide Nukada, respectively. Mouse PIP5K (48) was kindly provided by Tibor Rohacs (Rutgers New Jersey Medical School, NJ). The pAnap plasmid was obtained from Addgene (Addgene ID: 48696) (49).

Oocyte Preparation.

Oocytes were harvested from Xenopus laevis as previously described (1, 4) in accordance with the guidelines of the Animal Care and Use Committee of the University of Osaka Graduate School of Medicine. Complementary RNA (cRNA) was transcribed in vitro using a mMESSAGE mMACHINE kit (Thermo Fisher Scientific, USA) and 50 nL of cRNA was injected into defolliculated oocytes. The oocytes were maintained in ND96 solution (96 mM NaCl, 2 mM KCl, 1.8 mM CaCl₂, 1 mM MgCl₂, 5 mM HEPES, pH 7.5 with NaOH) supplemented with 0.1 mg/mL gentamycin (Nacalai Tesque, Inc., Japan) and 5 mM sodium pyruvate at 18 °C until recording.

TEVC of Xenopus Oocyte.

The TEVC was performed using an Oocyte Clamp amplifier (OC-725C: Warner Instruments, USA) at room temperature (20 to 26 °C). The glass microelectrodes were filled with the solution containing 2 M potassium acetate and 1 M KCl or the solution containing 3 M potassium acetate and 10 mM KCl, and their resistances ranged from 0.1 to 1.0 MΩ. To evaluate phosphatase activity of Ci-VSP, WT or Y255 mutant Ci-VSP was coexpressed with GIRK2d, Gβ, and Gγ on the same oocyte. Output signals were digitized at 10 kHz through an AD/DA converter (InstruTECH LIH 8+8: HEKA Elektronik, Germany) running under PatchMaster software (HEKA Elektronik). Sensing current was also measured from the same oocyte to assess expression level of Ci-VSP. Bath solution was ND96 solution and holding potential was at −60 mV.

To evaluate the Q_OFF-V relationship, sensing current was measured in the solution containing 100 mM NMDG, 2 mM KCl, 5 mM MgCl₂, and 5 mM HEPES, pH 7.6 with HCl. Sensing currents were evoked by depolarizing steps from the holding potential of −80 mV to 200 mV in 10-mV increments. Leak subtraction was performed using a P/-8 protocol.

VCF.

Ci-VSP and Ci-VSPTEN16h harboring 3-(6-acetylnaphthalen-2-ylamino)-2-aminopropanoic acid (Anap: FutureChem Co., Ltd., Korea) in the target site which was mutated to TAG were expressed in oocytes as described previously (3).

Anap fluorescence measurement was done using two different setups: an inverted fluorescence microscope (model IX71: Olympus, Japan) equipped with a 20x 0.75 N.A. objective (Olympus) and a 100-W mercury arc ramp (Olympus), and an upright fluorescence microscope (model BX50WI: Olympus) equipped with a 20x 1.0 N.A. water immersion objective (Olympus) and a 365 nm LED (M365LP1: Thorlabs, Inc., USA). The TEVC was done using an Oocyte Clamp amplifier (OC-725C) at room temperature (22 to 24 °C). The glass microelectrodes were filled with the solution containing 2 M potassium acetate and 1 M KCl or the solution containing 3 M potassium acetate and 10 mM KCl, and their resistances ranged from 0.1 to 1.0 MΩ. Fluorescence excitation of Anap, emission of the fluorescence from oocytes, and detection of the emitted fluorescence by two photomultiplier tubes (H10722-20: Hamamatsu Photonics, Japan) were conducted as previously described (21, 22, 50). The output signals from the photomultiplier tubes were digitized at 10 kHz with an AD/DA converter (Digidata 1440A) and analyzed using pClamp software. ND96 solution was used as the bath solution. Fluorescence changes were elicited by single 500-ms depolarization from the holding potential of −60 mV to 160 mV. This was repeated 3 to 10 times every 60 s (long interval protocol) and 0.7 s (short interval protocol) in the same oocyte. Acquired fluorescence data were averaged. For F_ss-V relationship, fluorescence changes were measured upon depolarizing steps from the holding potential of −60 mV to 180 mV in 20-mV increments with 0.7-s sweep interval. For F_peak-V relationship of Y255Anap and K555Anap, fluorescence changes were measured upon depolarizing steps from 40 mV (K555Anap) or 80 mV (Y255Anap) to 180 mV in 20-mV increments with 60-s sweep interval. In most part of this paper, Anap fluorescence within the 460- to 510-nm bandwidth was presented. That within the 420- to 460-nm bandwidth did not show a significant difference in the fluorescence time courses dependent on PI(4,5)P₂ level (between the long and the short interval protocols; SI Appendix, Fig. S11) except for A242Anap which showed the PI(4,5)P₂-dependent fluorescence change of the time course in this bandwidth but not in the 460- to 510-nm bandwidth (SI Appendix, Fig. S4C). For N251Anap, R254Q/Y255Anap, and R254G/Y255Anap, fluorescence within the 420- to 460-nm bandwidth was used for data analysis because of a weak signal within the 460- to 510-nm bandwidth (SI Appendix, Figs. S4C and S10C). For Q_OFF-V relationship, sensing currents were evoked by depolarizing steps from the holding potential of −60 mV to 160 mV in 20-mV increments. Leak subtraction was performed using a P/-8 protocol.

VCF with tetramethylrhodamine-6-maleimide (TMRM: Chemodex Ltd., Switzerland) was conducted using oocytes expressing Ci-VSP in which Q208 was substituted for a cysteine. The oocytes were labeled with TMRM as described previously (3).

The upright fluorescence microscope setup (model BX50WI) equipped with a 20x 1.0 N.A. water immersion objective and a cold white LED (MCWHL8: Thorlabs, Inc.) was used for TMRM fluorescence measurement. Fluorescence excitation of TMRM and emission of the fluorescence from oocytes were conducted as previously described (22). The emitted fluorescence was detected by one photomultiplier tube. Fluorescence changes were elicited by voltage steps from the holding potential of −60 mV to 200 mV in 10-mV increments. The set of step pulses was repeated three times, and acquired fluorescence data were averaged.

MD Simulations.

The predicted structure of Ci-VSP (3) missing the N-terminal intracellular flexible linker (M1 to T99) was used as an initial model for MD simulations. The MD simulation system was set up using the CHARMM-GUI web server (51). The lipid bilayer was composed of 1-palmitoyl-2-oleoyl-sn-glycero-3-phosphocholine (POPC), and approximately 20% and 5% concentrations of cholesterol were included on the cytoplasmic and the extracellular sides of the membrane leaflet, respectively. The concentration of PI(4,5)P₂ on the cytoplasmic side of the membrane leaflet was determined depending on the situation and ranged from 5 to 15%. The VSD was properly embedded in the lipid bilayer using the prediction from the Orientations of Protein in Membrane (OPM) database (52). The Ci-VSP-bilayer complex was solvated by TIP3P model water (53) under the periodic boundary condition, and then water molecules were replaced with an appropriate number of potassium and chloride ions to neutralize the system at the ionic strength of 150 mM. Electrostatic interactions were calculated using the Particle Mesh Ewald method (54). Chemical bonds associated with the hydrogen atom other than water were treated as rigid using the LINCS algorithm (55). The SETTLE algorithm (56) was used to keep the water model rigid. The CHARMM force field (57) was used for the protein, lipids, and ions. The time step was set to 2 fs. The total energy of the prepared system was minimized using the steepest descent method. The system was then equilibrated with a constant number of atoms, a fixed volume, and a temperature of 310 K (NVT). The system was further equilibrated with a constant number of atoms and a pressure of 1 bar at 310 K (NPT). These equilibrations were performed using the Berendsen barostat and thermostat (58). For the production run, we used the Parrinello–Rahman barostat (59) and the Nosé-Hoover thermostat (60, 61), saving snapshot structures at appropriate intervals from the trajectories for the analysis. MD simulations were performed by using GROMACS program (62). Each simulation was run longer than 250 ns.

Simulations were visualized using VMD (63). Figures were prepared using VMD and PyMOL (Schrödinger, Inc., USA).

Data Analysis.

Data are presented as the mean ± SD. Data analysis was performed using Microsoft 365 Excel, PatchMaster, Clampfit (Molecular Devices), and Igor Pro software. Statistical tests were made using Microsoft 365 Excel.

Phosphatase activity of VSP was taken as the rate constant which was determined by fitting of a normalized K_ir current decay with a single exponential function (3).

Supplementary Material

Appendix 01 (PDF)

Data, Materials, and Software Availability

All study data are included in the article and/or SI Appendix.