In this study, the phosphatase activity of Ci-VSP was assessed at four membrane potentials representing the entire voltage range over which the movement of the VSD varied; changes of K+ channel activity and PHPLC-GFP fluorescence served as two independent indexes of Ci-VSP activity at high membrane potentials. Measurement of Kir2.1 currents in both TEVC and on-cell patch configurations showed that the phosphatase activity increased as membrane potential became more positive (Figs 1 and 4). Similarly, phosphatase activity estimated from measurements of PHPLC-GFP fluorescence using a mathematical model revealed that Ci-VSP activity increased in the voltage range extending from 50 mV to 125 mV (Figs 9 and 12). Taken together with our earlier findings on phosphatase activity at lower membrane potentials (Murata & Okamura, 2007), the present study indicates that Ci-VSP phosphatase activity is coupled to the movement of the VSD over the entire range of its voltage sensitivity.
Validation of modelling of the PHPLC-GFP fluorescence and Kir2.1 currents
In our modelling of PHPLC-GFP levels, the effect of its diffusion away from the plasma membrane after PI(4,5)P2 depletion was incorporated into the Virtual Cell framework to calculate the density of phosphoinositides and PHPLC-GFP on the plasma membrane. By taking diffusion of PHPLC-GFP into account, a time course similar to that of the fluorescence signal was reconstituted by the model. In our model, the diffusion coefficient for PHPLC-GFP was estimated to be 4–12 μm2 s−1 (Figs 7 and 10D). The kinetics of the fluorescence signal simulated with a value of 45 μm2 s−1 (the right panel of Fig. 7D) did not match well with the experimental data of PHPLC-GFP. The diffusion coefficient of the PH domain fused to GFP in the cytoplasm has been studied in Dictyostelium using fluorescence correlation spectroscopy and it was estimated to be 11–27 μm2 s−1 (Ruchira et al. 2004), a range close to that applied in our analysis.
Besides diffusion of PHPLC-GFP polypeptide, the level of Ci-VSP expression influenced the apparent voltage dependence of the phosphatase activity (Fig. 2). When Ci-VSP was expressed at a higher level, voltage-induced changes in PHPLC-GFP fluorescence saturated at a voltage lower than the voltage at which there was no further increase in charge movement in the Q–V curve (Figs 6G and 8A). Incorporating the diffusion of PHPLC-GFP underneath the plasma membrane and the surface density of Ci-VSP into the model, calculated enzymatic activities showed similar profiles of voltage dependence based on the analyses of different oocytes with distinct expression level of Ci-VSP.
In contrast with the model of PHPLC-GFP fluorescence, a maximum R value of 0.3–0.4 was necessary to be used to represent the extent of the enzymatic activity (Fig. 10B), where the Vmax of Ci-VSP in oocytes is assumed to be 0.3 to 0.4 times that estimated using an in vitro phosphatase assay (Murata et al. 2005). Modelling with larger R values did not provide a good fit to the experimental results (data not shown). In modelling Kir current results, we chose a scheme where four PI(4,5)P2s are needed to bind Kir channel for the channel to be available. Modelling with one for this number of PI(4,5)P2 did not provide good fits of experimental results (data not shown), consistent with reports with tandem subunits with mutation of altered PI(4,5)P2 sensitivity where four PI(4,5)P2 are needed for full availability of Kir channels (Jin et al. 2008; Xie et al. 2008). On the other hand, recent reports (Jin et al. 2008; Xie et al. 2008) suggested that Kir channels with less than four PI(4,5)P2 are active to some extent, whereas in our modelling analysis these Kir channels were assumed to be silent. Despite such possible inaccuracy in our assumption of relationship between PI(4,5)P2 and Kir2.1 channel, setting the maximum R value to 0.3–0.4 still gave a good fit to the experimental data of PHPLC-GFP fluorescence, when the diffusion coefficient was changed to 6–12 μm2 s−1 (n= 4) (Fig. 10D). In addition, in the analysis of the data summarized in Figs 8 and 9, the reconstituted kinetics were not significantly altered by such modification of the parameters (Supplementary Fig. 2).
In the TEVC recording of Kir2.1 channel, recoveries of PI(4,5)P2 level after its depletion due to Ci-VSP's activities occurred in two phases, which differs from previous observations in mammalian cells (Falkenburger et al. 2010). The early recovery phase was similar to that predicted by the model, but the later phase deviated from the predicted time course. This may indicate that our model misses factors that contribute to the slow recovery in Xenopus oocytes. For example, PI-5 kinase activity may depend on the PI(4,5)P2 level or may be down-regulated upon persistent depletion of PI(4,5)P2.
Interestingly, Kir current recovery followed different time courses in the TEVC and on-cell patch configurations. In the on-cell patch configuration, the Kir current could recover through both kinase-dependent production of PI(4,5)P2 and lateral diffusion of PI(4,5)P2 from areas of the membrane beyond the patch membrane, since depletion of PI(4,5)P2 occurs in a local area in this configuration. When the recovery of the Kir current was reconstituted based on the model with lateral diffusion of PI(4,5)P2 and a diffusion coefficient of 0.1 μm2 s−1, the kinetics were similar to the experimental results shown in Fig. 4G (data not shown). However, rapid recovery through lateral diffusion may not be expected in our experimental set-up. This is because the global level of PI(4,5)P2 could be reduced when the entire oocyte membrane, except the area under patch pipette, is held to about 0 mV, a voltage at which Ci-VSP is partially active. Further, the lateral diffusion coefficient for PI(4,5)P2 in Xenopus oocytes is unknown, but the coefficient reportedly ranges widely, from 0.8 μm2 s−1 in HEK cells (Golebiewska et al. 2008) to 0.00039 μm2 s−1 in atrial myocytes (Cho et al. 2005). In addition, Cho et al. (2006) showed that the speed of lateral diffusion is different in CHO cells than in HEK cells.
It is also known that PI(4,5)P2 is sequestered in specific membrane domains, such as lipid rafts (Martin, 2001; Johnson et al. 2008). Improved on-cell patch recordings of Kir currents in the presence of Ci-VSP or an orthologue may enable establishment of a useful model with which to study local PI(4,5)P2 dynamics that are involved in many physiological events.
Insights into operating mechanism and physiological roles of Ci-VSP
The voltage range over which Ci-VSP exhibits its phosphatase activity is similar to the range over which the VSD exhibits charge movement (Figs 9B and 12B). VSDs in voltage-gated ion channels regulate the activity of the pore domain, though it has been established that the voltage dependency of the VSD is less steep than that of the ion conductance (Bezanilla, 2000). Because voltage-gated K+ channels generally form tetramers and exhibit cooperativity among the four subunits (Bezanilla, 2000), the voltage dependency of the ion permeation is sharper than that of VSD movement. Our finding that phosphatase activity is coupled to VSD movement over the entire voltage range suggests that, unlike voltage-gated K+ channels, Ci-VSP is not subject to cooperativity among VSDs, which is consistent with the recent finding that Ci-VSP exists as a monomer in Xenopus oocytes (Kohout et al. 2008).
What are the mechanisms underlying the coupling between the VSD and the phosphatase domain at the single protein level? One possibility is that the Ci-VSP enzyme domain has only two states: active and inactive. In that case, maximum and constant phosphatase activity is attained only when the VSD is in a fully activated state, and the number of Ci-VSP molecules in the active state increases as the membrane potential becomes more positive. The other possibility is that the enzymatic activity of single Ci-VSP proteins can be gradually tuned as the membrane potential is altered. Distinct levels of phosphatase activity could be coupled to distinct movement states of the voltage sensor. Further studies will be necessary to determine the precise mechanism of the functional coupling between the VSD and the phosphatase activity of Ci-VSP.
Our conclusion that Ci-VSP phosphatase activity is coupled to the VSD over a span of 100 mV also raises a simple question: does Ci-VSP function at such high membrane potentials in any physiological context? One recent study of Ci-VSP gene expression in ascidian juveniles and adults showed that Ci-VSP is expressed in a subpopulation of blood cells that express the gp91 NADPH oxidase subunit gene (Ogasawara et al. 2011). The gp91 subunit plays a principal role in generating reactive oxygen species in phagocytes, where large depolarizations (>50 mV) take place due to the export of electrons (Bánfi et al. 1999). In addition, phosphoinositides play critical roles in the dynamics of the generation and maturation of phagosomes and endosomes within phagocytes (Kamen et al. 2007; Tiwari et al. 2009). It will be interesting to see whether the voltage range of Ci-VSP's enzymatic activity is, in fact, important for activation in such native environments.