Modulation of High-Voltage Activated Ca²⁺ Channels by Membrane Phosphatidylinositol 4,5-Bisphosphate

SUMMARY

Modulation of voltage-gated Ca²⁺ channels controls activities of excitable cells. We show that high-voltage activated Ca²⁺ channels are regulated by membrane phosphatidylinositol 4,5-bisphosphate (PIP₂) with different sensitivities. Plasma membrane PIP₂ depletion by rapamycin-induced translocation of an inositol lipid 5-phosphatase or by a voltage-sensitive 5-phosphatase (VSP) suppresses Ca_V1.2 and Ca_V1.3 channel currents by ~35%, and Ca_V2.1 and Ca_V2.2 currents by 29 and 55%, respectively. Other Ca_V channels are less sensitive. Inhibition is not relieved by strong depolarizing prepulses. It changes the voltage dependence of channel gating little. Recovery of currents from inhibition needs intracellular hydrolysable ATP, presumably for PIP₂ resynthesis. When PIP₂ is increased by overexpressing PIP 5-kinase, activation and inactivation of Ca_V2.2 current slow and voltage-dependent gating shifts to slightly higher voltages. Thus, endogenous membrane PIP₂ supports high-voltage activated L-, N-, and P/Q- type Ca²⁺ channels, and stimuli that activate phospholipase C deplete PIP₂ and reduce those Ca²⁺ channel currents.

INTRODUCTION

Voltage-gated Ca²⁺ (Ca_V) channels mediate Ca²⁺ influx in response to membrane depolarization and regulate many physiological phenomena including neurotransmission, secretion, muscle contraction, and gene expression (Catterall et al., 2005). The activity of Ca_V channels is dynamically regulated by receptor-dependent signals, such as G proteins, protein kinases, calmodulin, soluble N-ethylmaleimide-sensitive fusion attachment receptor (SNARE) proteins, and the second messengers Ca²⁺ and arachidonic acid (Catterall, 2000; Dolphin, 2003; Roberts-Crowley et al., 2009). Here we analyze in detail the regulation of three high-voltage activated (HVA) Ca²⁺ channels (Ca_V1.2, Ca_V1.3, and Ca_V2.2) by the plasma membrane phospholipid phosphatidylinositol 4,5-bisphosphate (PIP₂).

Signals from G-protein-coupled receptors (GPCRs) suppress N-type Ca_V2.2 channels through two pathways in sympathetic neurons (Hille, 1994). The “fast” pathway is voltage dependent, membrane delimited, and insensitive to the intracellular Ca²⁺ chelator BAPTA. The fast suppression is induced by activating receptors coupled to the pertussis toxin (PTX)-sensitive G proteins G_o and G_i. It can be relieved by applying large positive pulses (Bean, 1989; Lipscombe et al., 1989; Zamponi and Snutch, 1998), and is understood as direct voltage-dependent binding of G-protein βγ subunits to N-type (Ca_V2.2) and P/Q-type (Ca_V2.1) Ca²⁺ channels (Herlitze et al., 1996; Ikeda, 1996; Dolphin, 2003). By contrast, the “slow” pathway is voltage independent, insensitive to PTX, and sensitive to BAPTA. Activation of G_q-coupled receptors initiates the slow pathway (Bernheim et al., 1991; Delmas et al., 2005; Michailidis et al., 2007; Roberts-Crowley et al., 2009). While the fast and slow pathways both reduce the current appreciably, neither fully eliminates it. A phenomenologically similar slow pathway also produces inhibitory modulation of L-type Ca²⁺ channels and M-type (KCNQ) K⁺ channels by G_q-coupled receptors in sympathetic neurons (Mathie et al., 1992) and in reconstituted systems (Shapiro et al., 2000; Bannister et al., 2002). We have speculated that the underlying signaling for slow suppression of these Ca²⁺ and K⁺ channels might be the same (Bernheim et al., 1991; Mathie et al., 1992; Hille, 1994). Here we ask if some or all of the slow suppression of Ca²⁺ currents is due to receptor-mediated depletion of PIP₂ as is true for slow suppression of KCNQ K⁺ current (Suh and Hille, 2002; Zhang et al., 2003; Brown et al., 2007).

Currents in the Ca_V2 channel family can be modulated by exogenous manipulation of membrane phosphoinositides (see reviews Delmas et al., 2005; Michailidis et al., 2007). Wu et al. (2002) concluded that depletion of membrane PIP₂ underlies a significant rundown of Ca_V2.1 (P/Q-type) currents seen in inside-out excised patch experiments. They showed that application of PIP₂ antibody to the intracellular side of giant membrane patches accelerates the Ca_V2.1 current rundown, whereas a brief application of PIP₂ or Mg-ATP retards the rundown. However, unexpectedly, they also reported that applied PIP₂ reduced the currents. The “inhibitory” effect of PIP₂ was actually a strong positive shift of channel voltage dependence (by ~40 mV). It was antagonized by conditions that activated cyclic AMP-dependent protein kinase. In subsequent studies, Gamper et al. (2004) reported that Ca_V2.2 N-type channels are also regulated by PIP₂. Macroscopic current rundown was significantly slowed by the application of PIP₂ to the intracellular side of excised membrane patches from Xenopus oocytes. In sympathetic neurons, the current suppression during muscarinic receptor activation was attenuated and slowed by intracellular perfusion of short-chain DiC₈-PIP₂. When the N-type Ca²⁺ currents were inhibited by G_q-coupled receptor activation, current recovery was blocked by 50 μM wortmannin, to inhibit PI 4-kinase. Thus, they proposed that depletion of PIP₂ on the plasma membrane is the cause of the G_q receptor-mediated slow inhibition of N-type Ca²⁺ currents in sympathetic neurons. The PIP₂ hypothesis has never been tested for L-, R-, or T- type channels in any system. In contrast, others have attributed the slow receptor-mediated inhibition of both N-type and L-type Ca²⁺ channels to production of arachidonic acid (Liu and Rittenhouse, 2003; Liu et al, 2006; Roberts-Crowley et al., 2009). Thus whether PIP₂ depletion is a major physiological signal for slow receptor-mediated suppression of N- and L-type Ca²⁺ channels remains controversial (Michailidis et al., 2007).

G_q-coupled receptor signals are notoriously difficult to dissect because they produce so many downstream second messengers. For example, in studies of G_q modulation of Ca²⁺ channels, PIP₂ depletion occurs simultaneously with the downstream production of arachidonic acid, activation of protein kinase C, and elevation of cytoplasmic free Ca²⁺, which are all believed to have significant effects on the channels. In order to test the PIP₂ hypothesis unambiguously, we here use two strategies to deplete PIP₂ enzymatically and rapidly without producing the downstream products of PLC. We take advantage of two exogenous polyphosphoinositide 5-phosphatase systems that can convert PI(4,5)P₂ directly to PI(4)P in the plasma membrane in living cell systems without activation of receptors. One system uses “chemical dimerization,” and the other uses membrane depolarization to activate transfected 5-phosphatase enzymes that convert PIP₂ to PIP. In chemical dimerization, addition of rapamycin or its analogue iRap to the extracellular medium induces irreversible translocation of a transfected yeast INP54p 5-phosphatase from cytosol to membrane, initiating PIP₂ dephosphorylation (Suh et al., 2006; Varnai et al., 2006). The translocation and dephosphorylation take 10 - 20 s. The other system uses a transfected voltage-sensitive phosphatase (VSP), an integral plasma membrane protein that becomes active when its N-terminal voltage-sensor domain detects large membrane depolarization (Murata et al., 2005; Halaszovich et al., 2008; Okamura et al., 2009). We find that the VSP system is faster, depleting plasma membrane PIP₂ within a second of activation, and the enzyme turns off quickly when the membrane is repolarized.

Here, we focus on the PIP₂ hypothesis and reserve examination of other messengers for later work. We consider N-type Ca²⁺ channels, which together with P/Q-type channels were the subject of the previous studies (Wu et al. 2002; Gamper et al. 2004; Lechner et al., 2005), and we consider two L-type channels whose PIP₂ dependence has not been studied before. In addition we screen the other subtypes of Ca_V channels. We find that PIP₂ depletion attenuates both L- and N-type Ca²⁺ channel activity through voltage-independent pathways, and increases the susceptibility of the channels to voltage-dependent inactivation (VDI). We also find that resynthesis of PIP₂ from PIP is needed for Ca_V channel recovery. Our experiments show that by itself PIP₂ depletion does depress HVA Ca²⁺ channel activity in living cells.

RESULTS

Our goal was to test the hypothesis that Ca²⁺ channels respond to depletion and resynthesis of PIP₂ in living cells. To establish conditions for recording Ca²⁺ channel currents, we expressed Ca_V1.3 (α1D) or Ca_V2.2 (α1B) channel subunits together with β3 and α2δ1 accessory subunits in tsA cells and recorded the whole-cell currents with Ba²⁺ as the charge carrier. Depolarizing voltage steps from a holding potential of −70 mV evoked inward currents carried by Ba²⁺ (Figure 1A). Barium was used to minimize Ca²⁺-dependent inactivation of the current so that any current decay observed during the test pulses would be due primarily to voltage-dependent inactivation (VDI). As expected, the Ca_V1.3 currents inactivated little during the 40-ms test pulses, whereas the Ca_V2.2 currents inactivated more. Peak current-voltage (I-V) relations showed that Ca_V1.3 currents peaked near −10 mV (n = 6), whereas Ca_V2.2 currents peaked near +10 mV (n = 8) (Figure 1B). These respective peak voltages were used for all test pulses in subsequent experiments to assay the function of these channels, except where indicated.

Figure 1. PIP₂ Depletion Depresses Ca_V1.3 and Ca_V2.2 Voltage-Gated Ba²⁺ Currents.

(A) Families of whole-cell Ba²⁺ currents elicited by voltage steps from −80 to +50 mV in 10-mV intervals (see pulse protocol), in cells expressing Ca_V1.3 (left) and Ca_V2.2 (right) channels. Holding potential is −70 mV and dashed line is zero current. Closed arrowheads indicate peak inward Ba²⁺ currents triggered by the depolarizing test pulses. Tail currents are clipped.

(B) Peak current-voltage (I-V) relations for Ca_V1.3 and Ca_V2.2 currents in whole-cell recording normalized to the maximum current. Points are mean ± SEM (Ca_V1.3, n = 6; Ca_V2.2, n = 8).

(C and D) Current modulation by iRap (5 μM) and Oxo-M (10 μM) in cells co-expressing M₁ muscarinic receptors and CF-Inp alone (top, −LDR) or with LDR (bottom, +LDR). Currents were recorded in response to test pulses to −10 mV (Ca_V1.3) or +10 mV (Ca_V2.2) every 4 s. Dashed line is zero current. LDR, membrane anchor protein. CF-Inp, PI 5-phosphatase.

(E and F) Percent inhibition of Ca_V1.3 (E) and Ca_V2.2 (F) currents by iRap and Oxo-M compared to initial currents in cells expressing CF-Inp alone (−LDR) or with LDR (+LDR). The application of iRap significantly inhibited Ca_V1.3 currents (*P < 0.05 compared to iRap effect without LDR, n = 3 for −LDR; n = 4 for +LDR) and Ca_V2.2 currents (**P < 0.01 compared to iRap effect without LDR, n = 3 for both −LDR and +LDR). Data are mean ± SEM. See also Figure S1.

Chemical Translocation of a PIP₂ 5-phosphatase to the Plasma Membrane Attenuates Ca_V Currents

We begin with the chemical dimerization system and iRap to deplete membrane PIP₂. In addition to the channel subunits and the M₁ muscarinic receptor, two additional components needed to be co-transfected: the membrane-localized iRap-binding protein Lyn₁₁-FRB (LDR) and the fluorescent cytoplasmic enzyme construct CFP-FKBP-INP54p (CF-Inp). The CF-Inp has PIP₂ 5-phosphatase activity that converts PIP₂ to PI(4)P. We showed previously that when LDR and CF-Inp are brought together at the membrane by application of iRap, PIP₂ is irreversibly depleted and PIP₂-dependent KCNQ K⁺ channels turn off (Suh et al., 2006). This system is suitable for experimental designs that benefit from the irreversible PIP₂ depletion that follows chemical dimerization.

As an initial control, when CF-Inp is expressed but the membrane anchor LDR is omitted (−LDR), iRap had little direct effect on the currents (top panels in Figures 1C and 1D), although both channels were readily inhibited by M₁ receptor activation with the muscarinic agonist Oxo-M. When LDR was included (+LDR) there were two changes. First, currents in Ca²⁺ channels were decreased irreversibly by iRap (bottom panels in Figures 1C and 1D), a decrease that was less than had been seen with muscarinic receptor activation. The second effect was alteration of the subsequent response to muscarinic receptor activation. Prior PIP₂ depletion by activated INP54p 5-phosphatase eliminated further inhibition of Ca_V1.3 current by muscarinic receptors (Figure 1C), quite possibly because the irreversible depletion of PIP₂ abrogates muscarinic generation of all PIP₂ cleavage products, including inositol trisphosphate and calcium signaling (Suh et al., 2006). On the other hand, further muscarinic modulation of Ca_V2.2 current remained intact and reached full amplitude (Figure 1D). Quite possibly that pathway does not require PIP₂. It might involve G protein βγ subunits or other products of phospholipases including PLA₂ (Melliti et al., 2001; Roberts-Crowley et al., 2009). As a preliminary conclusion, muscarinic inhibition of both channels probably occurs via more than one pathway, and any PIP₂ depletion component accounts for only part of the total effect.

It is well known that inhibition of Ca_V2.2 current by M₂ muscarinic receptor-mediated signaling (the fast Gβγ pathway) is strongly relieved by large positive prepulses (Elmslie, et al., 1990). We readily verified this effect in cells transfected with M₂ (G_i-coupled) rather than M₁ (G_q-coupled) receptors (data not shown). Can the iRap-induced inhibition of Ca_V1.3 and Ca_V2.2 channels also be relieved by strong depolarizing prepulses? The cells were given a +130-mV/20-ms prepulse followed 5 ms later by a 10-ms test pulse to measure channel function. Figure S1 shows that inhibition was unchanged by the prepulses for Ca_V1.3 channels (Figure S1A) and for Ca_V2.2 channels (Figure S1B). Hence, unlike inhibition of Ca_V2.2 channels via Gβγ, positive prepulses do not relieve the suppression that follows iRap-induced PIP₂ depletion.

Depletion of Membrane PIP₂ by Activation of Dr-VSP Attenuates Ca_V Currents

We turn now to depleting PIP₂ with the voltage-sensitive phosphatase from zebra fish (Dr-VSP; Okamura et al., 2009). This tool is suitable for experimental designs that benefit from reversible PIP₂ depletion following an activating depolarization. First we characterized the ability of Dr-VSP to deplete membrane PIP₂ by using two fluorescent PIP₂ indicators and measuring fluorescence resonance energy transfer (FRET) between them (van der Wal, 2001; Jensen et al. 2009). In resting cells, the PIP₂-binding fluoroprobes CFP-tagged PH(PLCδ1) and YFP-tagged PH(PLCδ1) bind to the PIP₂ at the plasma membrane in high enough surface density to generate FRET between them (Figure S2A). If PIP₂ is depleted, the probes dissociate from the membrane and move to the cytosol, losing their FRET interaction (van der Wal et al., 2001). In cells cotransfected with Dr-VSP, application of depolarizing pulses to +120 mV activated the phosphatase activity. Using 440-nm light to excite CFP, the fluorescence of CFP (CFP_C) increased and that of YFP (YFP_C) decreased each time the depolarization was applied (Figure S2B, top), and PIP₂ depletion was signaled as the corresponding decrease in the FRET ratio (FRET ratio = YFP_C/CFP_C) (Figure S2B, bottom). During the large depolarization, the PIP₂ depletion developed rapidly (exponential time constant τ = 105 ± 18 ms, n = 9) (Figure S2C) and in a voltage dependent manner (V_1/2 = 61 ± 5 mV, n = 5) (Figure S2D). According to the FRET assay, the depletion with Dr-VSP is comparable to that seen with M₁ muscarinic receptor activation (Figure S2E). However, it is much faster and results in quite different cleavage products.

As was anticipated from the PIP₂ depletion, activation of Dr-VSP decreased current in HVA Ca²⁺ channels. Our protocol was to apply a standard test pulse (pulse a) to record baseline channel current, then a large depolarizing pulse for various times to activate Dr-VSP, followed by a second test pulse (pulse b) (Figure 2A). In control cells not expressing Dr-VSP, Ca_V1.3 current amplitudes a and b were almost the same without (0 s) or with (0.5 s) a 0.5-s depolarization to +120 mV (Figure 2A, left). In contrast, in cells expressing Dr-VSP, the 0.5-s depolarizing pulse significantly attenuated the Ba²⁺ current in pulse b (Figure 2A, middle). Again, in the same cell there was no significant change in current b without the large pulse. To examine whether this Dr-VSP-induced inhibition of Ca_V1.3 current is caused by PIP₂ degradation, we tested the effect of Dr-VSP activation in cells transfected with the PIP 5-kinase type-1γ (PIPKIγ). This enzyme elevates PIP₂ concentration in the plasma membrane (Wenk et al., 2001) and thereby diminishes the ability of G_q-coupled receptors to suppress KCNQ K⁺ currents (Suh and Hille, 2007). As shown in Figures 2A and 2C right, the inhibition of Ca_V1.3 and Ca_V2.2 channels by Dr-VSP was significantly attenuated by PIPKIγ expression. Figure 2B plots the time dependence of the b/a current ratio (top) and of the percent inhibition with and without Dr-VSP (bottom). The Ca_V1.3 channels were maximally inhibited by 33% with an onset time constant τ = 112 ± 7 ms (Figure 2B, bottom left). Further, PIPKIγ overexpression significantly attenuated the Dr-VSP-induced Ca_V1.3 inhibition (Figure 2B, top left).

Figure 2. Inhibition of Ca_V Currents by Dr-VSP-Mediated PIP₂ Depletion.

(A and C) Typical traces of Ca_V1.3 (A) and Ca_V2.2 (C) currents before and after activation of Dr-VSP by depolarizations to +120 mV. Cells without Dr-VSP (Control), cells transfected with Dr-VSP, or cells transfected with Dr-VSP plus PIPKIγ received a test pulse to −10 (A) or +10 mV (C) for 10 ms and then were depolarized to +120 mV for zero or 0.5 s (as marked), followed by a second test-pulse. The currents before (a) and after (b) the +120 mV-depolarizing pulse are superimposed. Dashed line is zero current.

(B and D) Time-dependent inhibition of L-type (Ca_V1.3, B) and N-type (Ca_V2.2, D) currents by Dr-VSP activation. Top, cells were depolarized to +120 mV for various times and the relative current ratio (b/a) was measured in control (open circle, n = 6-14 for Ca_V1.3 and n = 6 for Ca_V2.2), Dr-VSP-expressing (closed circle, n = 6-11 for Ca_V1.3 and n = 6 for Ca_V2.2) cells, and Dr-VSP plus PIPKIγ expressing cells (n = 5-8). The delay between subsequent test pulses was 1 min. Bottom, percent inhibition of currents by time-graded activation of Dr-VSP at +120 mV. See formula in text. The current inhibition by 1-s depolarizing pulse is labeled in each figure. See also Figure S2.

We also performed Dr-VSP experiments with N-type Ca_V2.2 channels. The bottom line was similar: Ca_V2.2 channels were inhibited by Dr-VSP activation with a maximum inhibition of 56% and onset τ = 143 ± 10 ms (Figure 2D, bottom). However, in these experiments there was much more evidence of channel inactivation induced by the voltage protocols. Even in the absence of Dr-VSP, the Ca_V2.2 current in pulse b was reduced by as much as 31 ± 2% (n = 6) by the preceding 0.5-s depolarization to +120 mV (Figures 2C and D). Moreover, N-current was even partially reduced without (0 s) the large depolarizing pulse. To compensate for such “control” inactivation, we calculated the percent inhibition due to Dr-VSP action by the formula 100 {1 - (b/a)_VSP/(b/a)_Control} at each time point in this figure and in subsequent figures. This correction would apply accurately if the “inactivation” seen without Dr-VSP is unchanged by the action of Dr-VSP, an assumption that we revisit later. Hence, Dr-VSP, like the iRap-dimerizable INP54p system, depletes PIP₂ and leads to parallel depression of currents in three subtypes of Ca²⁺ channels. We did find that, for Ca_V1.3 and Ca_V2.2 channels, the mean inhibition by Dr-VSP was ~35% larger than that with the iRap system. In summary, experiments with two lipid phosphatases are consistent with the hypothesis that PIP₂ regulates Ca_V channels.

Inhibition of Ca_V Channels by Dr-VSP Is Not Simple VDI

We have already said that much of the current inactivation developing during test pulses in the presence of barium is voltage-dependent, VDI. In Figure 3 we examined the role of VDI in our measurements and its possible dependence on PIP₂. We ask whether some parts of the inhibition by Dr-VSP are simply an enhancement of VDI by testing whether the inhibition can be removed by a large hyperpolarizing pulse. The 0.9-s hyperpolarizing voltage step to −150 mV ended 0.1 s before test pulse b (see pulse protocols in Figures 3A and 3C).

Figure 3. Channel Inhibition by Dr-VSP Corrected for VDI.

(A and C) Effect of membrane hyperpolarization on Dr-VSP-induced inhibition of Ca_V1.3 (A) and Ca_V2.2 (C) currents. The changes of Ca_V1.3 and Ca_V2.2 currents by a +120-mV depolarizing pulse were measured without (−) and with (+) a hyperpolarizing step (−150 mV, 0.9 s) in control and Dr-VSP-expressing cells. Pairs of current traces were recorded from the same cell with a 1-min interval.

(B and D) Summary of the relative peak current (b/a) of Ca_V1.3 (B) and Ca_V2.2 (D) in control and Dr-VSP-expressing cells with and without the hyperpolarizing step. Data are mean ± SEM (Ca_V1.3, n = 5 - 6; Ca_V2.2, n = 4). The percent difference between control and Dr-VSP is labeled in each condition.

(E) Effect of Dr-VSP on Ca_V2.2 tail currents. Tail currents were measured with a hyperpolarizing step (−150 mV, 0.4 s) in control and Dr-VSP-expressing cells. Pairs of current traces were recorded from the same cell 1 min apart. Capacitive and leak currents were subtracted by a P/5 procedure.

(F) Summary of the tail-current inhibition (%) of Ca_V2.2 in control and Dr-VSP-expressing cells. Data are mean ± SEM (control, −6.1 ± 7.2, n = 5; Dr-VSP, 57.6 ± 5.2, n = 8).

In control cells, Ca_V1.3 currents showed no or very minor VDI from the +120-mV/1-s depolarizing pulse, i.e., in control cells, currents a and b were very similar even without the hyperpolarization. With the −150-mV hyperpolarizing voltage step, the minor VDI was abolished (Figure 3A, top right). In cells expressing Dr-VSP, Ca_V1.3 current b was strongly reduced compared to a, without and with the hyperpolarizing step (Figures 3A, bottom, and 3B). Thus, Ca_V1.3 channels had little residual VDI from our pulse protocol, and the inhibitory effect of Dr-VSP activation was not relieved by hyperpolarizations. In control cells expressing Ca_V2.2 channels, there was some residual VDI after pulse a (Figure 3C, top left). This made the pulse b currents ~30% smaller (Figures 3C and 3D). The reduction was totally relieved by the −150-mV/0.9-s hyperpolarization. Indeed, the current in pulse b became larger than that in pulse a as if the hyperpolarization was also relieving some resting inactivation that had reduced pulse a current (Figure 3C, top right). With Dr-VSP, the large depolarizing pulse strongly depressed current in pulse b as before. Again when compared to the control cells, the hyperpolarizing step did not relieve any of the effect of VSP (Figure 3D). Ca_V2.2 channels show prominent tail currents. Therefore, we also could test whether the tail currents were inhibited by Dr-VSP. Figures 3E and F show strong inhibition in Dr-VSP expressing cells. The inhibition of tail currents was similar to that of inward currents during test pulses. Thus for Ca_V1.3 and Ca_V2.2 channels, the depression of current due to Dr-VSP activation did not seem to be some kind of enhancement of VDI.

Subtype-Specificity of PIP₂ Modulation of Ca_V Channels

We screened for channel modulation by expressing different α₁ subunits with the same β3 and α2α1 channel subunits as before. Figure 4A shows that the Ca_V1.2 and Ca_V2.1 channel currents were also significantly inhibited by Dr-VSP activation although less than for Ca_V2.2 channels. Since the current-voltage relations for Ca_V1.2 and Ca_V2.1 peaked at +10 mV and 0 mV, respectively (Figure S3A), those voltages were used for the test pulses in these experiments. The Ca_V1.2 channels gave currents similar to those with Ca_V1.3 and showed almost the same inhibition by muscarinic activation (62 ± 8% for Ca_V1.2, n = 5; 59 ± 3% for Ca_V1.3, n = 6) (Figures 4B and 4C, bottom). Furthermore, the effects of Dr-VSP activation were very similar (Figure 4C, top). With Ca_V1.2 channels, the large depolarizing pulse produced a 35% inhibition of current developing with an onset time constant τ = 138 ± 18 ms (n = 6). Ca_V2.1 channels are inhibited by both Dr-VSP and M₁ receptor stimulation (Figure 4). However, the other subtypes of Ca_V channels, 1.4, 2.3, and all Ca_V3 (T-type), were insensitive to Dr-VSP activation (Figures 4C and S3B). Interestingly, the four PIP₂-depletion-sensitive channels were also strongly inhibited by M₁ receptor activation with Oxo-M, and the other channels were not significantly inhibited by Oxo-M. Indeed Ca_V2.3 R-type currents were strongly enhanced (Figure S3C).

Figure 4. Screening Ca_V Subtypes for Modulation by Dr-VSP and M₁ Muscarinic Receptors.

(A) Inhibition of Ca_V1.2 and 2.1 currents by Dr-VSP-induced PIP₂ depletion. Ca_V currents were measured during test pulses before and after a +120-mV/1-s depolarizing pulse in control and Dr-VSP-expressing cells. Typical current traces for each channel type are superimposed.

(B) Inhibition of Ca_V1.2 and 2.1 currents by M₁ muscarinic receptor stimulation with Oxo-M (10 μM). Insets, Typical traces before and after Oxo-M application were superimposed.

(C) Inhibition of Ca_V currents by Dr-VSP-induced PIP₂ depletion (top) and M₁ muscarinic receptor stimulation with Oxo-M (B). ND, not determined. *The Ca_V2.3 current was enhanced 2.7 ± 0.5-fold (n = 4) by the activation of M₁ receptors. See also Figure S4.

PIP₂ Dependent Modulation of Ca_V2.2 Channels

PIP₂-dependence of Ca_V current modulation was investigated in more detail with Ca_V2.2 channels. When the cells overexpressed the PIP₂-binding peptide scavenger, PH domain of PLCδ1, the current density in the transfected cells was significantly decreased compared to control cells expressing only GFP or cells expressing PH domain of Akt protein which binds to PI(3,4)P₂ and PIP₃ in the plasma membrane (Figure 5A). Next we tested if PIP₂ elevation above its normal level attenuates the muscarinic suppression of the Ca_V2.2 channels. Overexpression of PIPKIγ significantly attenuated the M₁ muscarinic receptor-induced inhibition (Figure 5B) as well as the Dr-VSP-induced Ca_V2.2 inhibition (Figure 2C). The muscarinic inhibition without PIPKIγ was 72 ± 8% (n = 6), and with PIPKIγ it was reduced to 35 ± 7% (n = 6).

Figure 5. PIP₂-Dependent Modulation of Ca_V2.2 N-type channels.

(A) Ca_V2.2 current density (pA/pF) was measured in cells expressing the Ca_V2.2 channels plus GFP, GFP-PH-PLCδ1, or YFP-PH-Akt. The cells were transfected with the same amounts of cDNA. Average membrane capacitances for cells are 22 ± 2 pF for GFP (n = 11), 25 ± 4 for PH-PLC (n = 11), and 24 ± 4 for PH-Akt (n = 12). Top, Confocal images of tsA cells expressing each fluorescent protein. Cell diameters were 20 - 30 μm.

(B) Elevated PIP₂ levels attenuate Ca_V2.2 channel inhibition by M₁ receptor stimulation. Ca_V2.2 currents were measured in control cells and in cells transfected with PIPKIγ. Right, Summary of current inhibition by Oxo-M. Data are mean ± SEM. *P < 0.05, compared to control.

(C) Current-voltage (I-V) relations of N-type Ca_V current in isolated rat SCG neurons expressing Dr-VSP (n = 3) with a widefield image of one neuron.

(D) Inhibition of N-type Ca_V currents by Dr-VSP activation in SCG neurons. N-type Ca_V currents were measured during test pulses before and after a +120-mV/1-s depolarizing pulse in control and cells expressing Dr-VSP. Capacitive and leak currents were subtracted by a P/5 procedure. Right, Summary of N-type current inhibition by Dr-VSP activation in SCG neurons. Data are mean ± SEM (n = 3 for control, n = 3 for Dr-VSP).

Finally, we tested if PIP₂ depletion could decrease the endogenous Ca_V2.2 N-type current of sympathetic superior cervical ganglion (SCG) neurons. Figure 5C shows the current-voltage relationship of N-type current in SCG neurons expressing Dr-VSP. The current peaked at ~10 mV. When the Dr-VSP was activated by +120 mV/1 s-depolarization, it inhibited the N-type current by 28% in SCG neurons (Figure 5D). Thus membrane PIP₂ is also important for Ca_V channel activity in differentiated neurons.

Recovery from Dr-VSP-Induced Inhibition Requires Intracellular ATP and PIP₂ Resynthesis

After the Dr-VSP-induced inhibition, current recovered in < 1 min. We tested the need for PIP₂ synthesis in channel recovery. Current was elicited with the voltage protocols shown in Figures 6A and 6E. Cells were given a 10-ms test pulse to measure the initial current (I₀), then depolarized to +120 mV for 1 s to activate Dr-VSP. Finally, recovery from inhibition was measured by applying 10-ms test pulses with successively longer delay after VSP activation starting at 0.5 s as indicated above the traces. As before, in control cells expressing Ca_V1.3 channels, the current was only slightly inhibited by the large depolarizing pulse (Figure 6A, top). With Dr-VSP, the Ca_V1.3 current was reduced by the depolarization and recovered to the initial level with a recovery time constant τ of 5.9 s (Figures 6A, bottom, and 6B). In control cells expressing Ca_V2.2 current, the current was reduced by VDI and then recovered (Figure 6E, top). In cells with Dr-VSP, the current was strongly inhibited by the large depolarization and, after correcting for control VDI, recovered with a time constant τ of 16.1 s (Figures 6E, bottom, and 6F), slower than for Ca_V1.3 channels. The slower recovery of Ca_V2.2 compared to Ca_V1.3 did not seem to be due to some form of slowly recovering VDI, since it was not significantly relieved or speeded by a −150-mV hyperpolarizing step (Figure S4A).

Figure 6. Recovery of Ca_V Currents after Dr-VSP-Induced Inhibition.

(A and E) Current traces for Ca_V1.3 (A) and Ca_V2.2 (E) channels in control (top) and Dr-VSP-expressing (bottom) cells before and after a +120-mV/1-s depolarizing pulse. Ca_V1.3 and Ca_V2.2 currents were measured at −10 mV and +10 mV, respectively, at the indicated times after the +120-mV pulse. Dashed lines indicate zero current, and dotted lines, the initial Ca_V current before the depolarization step.

(B and F) Time course of recovery of Ca_V1.3 and Ca_V2.2 currents after the Dr-VSP-induced inhibition in control (open circles) and Dr-VSP-expressing (closed circles) cells. Data are mean ± SEM (Ca_V1.3, n = 6 for both control and Dr-VSP; Ca_V2.2, n = 5 for both). Inset shows % current relative to control cells.

(C and G) Time course of Ca_V1.3 and Ca_V2.2 current recovery in cells transfected with PIPKIγ (Ca_V1.3, n = 5 for control and n = 8 for Dr-VSP; Ca_V2.2, n = 4 for control and n = 5 for Dr-VSP). Inset shows % current, comparing Dr-VSP to control cells.

(D and H) Time course of Ca_V1.3 and Ca_V2.2 current recovery with 3 mM AMP-PCP instead of ATP in the pipette solution. Data are mean ± SEM (Ca_V1.3, n = 7 for control and n = 10 for Dr-VSP; Ca_V2.2, n = 6 for control and n = 5 for Dr-VSP). Insets show the % current recovery in the presence of AMP-PCP. See also Figure S4.

We next tested the hypothesis that PIP₂ resynthesis is needed for Ca_V current recovery from the Dr-VSP-induced inhibition. First we speeded resynthesis. Coexpression of the 5-kinase PIPKIγ withCa_V1.3 or Ca_V2.2 channels significantly decreased the current inhibition with the +120-mV/1-s pulse and speeded the current recovery (Figures 6B, 6C, 6F, and 6G). Next we slowed PIP₂ resynthesis. The synthesis of PIP₂ from PI(4)P requires intracellular ATP, so we slowed the kinase activity by dialyzing the nonhydrolyzable ATP analogue AMP-PCP into the cell. The inclusion of 3 mM AMP-PCP instead of ATP in the pipette solution did not significantly affect maximum channel inhibition, but strongly slowed the recovery of both Ca_V1.3 current (τ = 21 s) (Figure 6D) and Ca_V2.2 current (τ = 32 s) (Figure 6H) and diminished the maximum recovery (for traces, see Figure S4B). As expected, dialyzing with AMP-PCP also strongly slowed and depressed PIP₂ resynthesis as measured by FRET with PH-domain probes (Figure S4C). With ATP, the Dr-VSP-induced FRET ratio changes recovered almost completely (94 ± 3%) with a time constant τ of 6.4 ± 0.9 s (n = 11), whereas with 3 mM AMP-PCP the recovery after one large depolarizing pulse was smaller (only 58 ± 3%) and slower with a time constant τ = 32 ± 4 s (n = 5), and there was no recovery after a second depolarizing pulse as if the last remaining ATP had been exhausted (Figure S4D). Inclusion of another nonhydrolyzable ATP analogue AMP-PNP gave a similar retardation of the FRET recovery (data not shown). In summary, we find that channel recovery after PIP₂ depletion is faster when PIP₂ synthesis is speeded and slower and incomplete when PIP₂ synthesis is slowed, implying that PIP₂ resynthesis underlies Ca_V channel recovery from the VSP-mediated inhibition.

Simultaneous Measurements of Channel Modulation and PIP₂ Degradation in the Same Cells

A puzzling finding was that recovery of Ca_V1.3 channels (Figure 6) closely paralleled that of FRET ratio measured in separate experiments (Figure S2), whereas recovery of Ca_V2.2 channels was slower than that of FRET ratio. Could it be that because we studied one set of cells expressing PH domains and different sets of cells expressing the channels, the comparison was not valid? It seemed necessary to cotransfect PH domains and channels and to measure the current and FRET ratio simultaneously in the same cell.

The following experiments show in simultaneous recordings, that the close parallels between Dr-VSP effects on Ca_V1.3 currents and FRET ratio changes persist. Figure 7A measures the onset of the Dr-VSP effects with depolarizations of different duration (Δt). The duration dependence was indistinguishable (Figure 7A, bottom), and the ratio of the time constants for onset (τ current /τ FRET) was 1.02 ± 0.09 (n = 5). Figure 7C shows the dependence on the voltage of the pulse for VSP activation; the mid-point voltage was 58 mV for both responses. Similarly, the recovery time courses of current and FRET ratio after termination of the depolarizing pulse were the same (time constant ratio 0.91 ± 0.09, n = 5; Figure 7D); when AMP-PCP replaced ATP in the pipette, the recoveries remained parallel although much slowed (Figure 7F). On the other hand, in similar simultaneous recording experiments with Ca_V2.2 channels, differences in time course persisted. The duration dependence for onset showed quicker loss of current than of FRET ratio (time constant ratio 0.65 ± 0.04, n = 4; Figure 7B) and the recovery after Dr-VSP showed slower recovery of current (time constant ratio 3.7 ± 0.7, n = 6; Figure 7E). These ratios ought to be interpreted cautiously since it was not possible to correct for confounding VDI in these experiments.

Figure 7. Simultaneous Measurement of Ca_V Current Modulation and PIP₂ Depletion in Single Cells.

All cells co-express channel subunits, PH-domain probes, and Dr-VSP. (A and B) Single-cell measurements of FRET ratio signals and whole-cell current from Ca_V1.3 (A) or Ca_V2.2 (B) channels. Top, Time-dependent induction of Dr-VSP effect on current and FRET ratio measured simultaneously in single cells. Bottom, superimposed time courses of current inhibition and FRET ratio decrease, normalized.

(C) Voltage dependence of Dr-VSP action on Ca_V1.3 current and FRET ratio change in a single cell.

(D and E). Time course of recovery of FRET ratio signals and whole-cell current of Ca_V1.3 (E) or Ca_V2.2 (F) channels in single-cell experiments. Top, recovery of currents and FRET ratio from the Dr-VSP-induced changes was measured simultaneously in single cells. Bottom, superimposed recoveries of current and FRET ratio in a single cell.

(F) AMP-PCP in the pipette solution attenuates the recovery of Ca_V1.3 current and FRET ratio. A single cell dialyzed with 3 mM AMP-PCP was given a 3-s or 5-s depolarizing pulse and current and FRET ratio were measured simultaneously.

Modulation of Ca_V2.2 Currents by Dr-VSP is not Gβγ binding

Ca_V2.2 (N-type, α1B) channel currents can be suppressed by membrane Gβγ subunits in a voltage-dependent manner (Dolphin, 2003). Might the Dr-VSP-mediated channel modulation be due in part to enhanced binding of Gβγ subunits to Ca_V2.2 channels when membrane PIP₂ is depleted? As a test we took advantage of the Gβγ subunit-insensitive chimeric channel construct called α1C-1B (Agler et al., 2005). In this construct, the N-terminus of Ca_V2.2 (N-type, α1B subunit), which includes one of the Gβγ binding sites, is replaced by the N-terminus of Ca_V1.2 (L-type, α1C subunit) (Figure 8A). When expressed in tsA cells, these chimeric channels activated in the same voltage range as wild type Ca_V2.2 channels but could not be inhibited by stimulation of M₂ (G_i-coupled) muscarinic receptors (Figures 8B and 8C). The wild type Ca_V2.2 channels were readily inhibited (78 ± 5%, n = 5). In cells expressing the chimeric channels and Dr-VSP, a +120-mV/1-s depolarizing pulse strongly inhibited the current by 56% when corrected for VDI seen in control cells (Figures 8D and 8E). Following the inhibition, the chimeric channels recovered with a time constant τ of 15 s (Figure 8F), comparable to the wild type channels (Figure 6F). These experiments give no evidence for enhanced binding of Gβγ subunits to the channel when PIP₂ is depleted. They also show that replacing the N-terminus of the Ca_V2.2 with that of Ca_V1.2 subunits does not change PIP₂-mediated channel modulation.

Figure 8. Modulation by Dr-VSP in a Gβγ-Insensitive Chimeric Ca_V2.2 Channel.

(A) Normalized peak current-voltage (I-V) relations of wildtype Ca_V2.2 (α1B) and chimeric Ca_V2.2 (α1C-1B) channels in the whole-cell configuration. Currents were elicited by voltage-steps from −40 to +40 mV, in 5 mV intervals, from a holding potential of −80 mV. Points are mean ± SEM (n = 5 for both channels).

(B) Time course of M₂ muscarinic receptor action (Oxo-M, 10 μM) on wild type Ca_V2.2 (α1B) (top) or Ca_V2.2 (α1C-1B) (bottom) channels. The current amplitude was measured at +10 mV every 4 s.

(C) Summary of the muscarinic inhibition of Ca_V2.2 (α1B) and Ca_V2.2 (α1C-1B) currents by M₂Rs. Data are mean ± SEM (Ca_V2.2 (α1B) alone, n = 5; Ca_V2.2 (α1B) with M₂ receptors, n = 4; Ca_V2.2 (α1C-1B) with M₂ receptor, n = 5).

(D) Inhibition of chimeric Ca_V2.2 (α1C-1B) currents by Dr-VSP. Typical traces of Ca_V2.2 (α1C-1B) currents before and after Dr-VSP activation by depolarization to +120 mV. Control (left) and Dr-VSP-expressing (right) cells received a test pulse and then were depolarized to +120 mV for zero or 1 s (Δt), followed by the second test pulse (b). The currents before (a) and after (b) the depolarizing pulse superimposed.

(E) Summary of the current inhibition (%) by the +120-mV depolarizing pulse in control (n = 8) and Dr-VSP-expressing (n = 5) cells.

(F) Time course of current recovery from Dr-VSP-induced inhibition. Cells were depolarized to +120 mV for 1 s, and the recovery of currents was measured. Inset shows the percent current from comparing control and Dr-VSP-expressing cells. Data are mean ± SEM (n = 6).

Does PIP₂ Depletion Change Channel Gating Properties?

A preliminary examination revealed only modest effects on channel gating as plasma membrane PIP₂ was depleted or raised. Some speeding of the development of VDI by PIP₂ depletion is shown in Figure 9A, again using our a/b pulse protocols but with longer test pulses. In control cells without Dr-VSP, the +120-mV/1-s depolarizing pulse had no effect on the time constant of VDI development during the subsequent 500-ms test pulse (Figure 9A, top panels). For Ca_V1.3 channels, the time constant (τ) of inactivation was 200 ± 27 s (n = 5) vs. 193 ± 22 s (n = 5), in pulses a and b, respectively, and for Ca_V2.2 channels, τ values were 44 ± 1 s vs. 42 ± 1 s (n = 5). However, with Dr-VSP, the time constant of inactivation was shortened, especially for Ca_V2.2 channels (49 ± 4 s vs. 25 ± 2 s for currents a and b, n = 5, *P < 0.001) (Figure 9A, bottom right). We note that test-pulse depolarizations to +10 mV produced no change in PH-domain FRET signals and thus do not activate Dr-VSP (Figure S2D). For Ca_V2.2 channels, we also explored effects of an increase in membrane PIP₂ levels by over-expressing the enzyme PIPKIγ. With elevated PIP₂, the development of VDI was slowed by ~1.7-fold at +10 mV and slowed by ~2-fold at +30 mV compared to control (Figures 9B and 9C). The activation of Ca_V2.2 channels was also slowed with expression of PIPKIγ, delaying the time to peak current (Figures 9D, S5A and S5B). Finally, effects on the voltage dependence of activation were small. When PIP₂ was depleted by the combination of Dr-VSP and AMP-PCP, the voltage dependence of activation of Ca_V1.3 channels was not changed and that for Ca_V2.2 channels showed a statistically insignificant left shift (Figure S5C and S5D). On the other hand, when PIP₂ was augmented by PIPKIγ, the current-voltage relation for Ca_V2.2 channels was significantly right shifted by 5-7 mV (Figure 9E). None of these small gating changes is sufficient to account for the large depression of currents that we have described following PIP₂ depletion. Rather it seems that with reduced PIP₂, fewer Ca_V channels are available to open.

Figure 9. PIP₂ Depletion Affects Channel Inactivation and Activation.

(A) Effect of Dr-VSP-induced PIP₂ depletion on the rate of inactivation of Ca_V1.3 (left) and Ca_V2.2 (right) currents. Ca_V currents were measured during 500-ms test pulses to −10 (Ca_V1.3) or +10 mV (Ca_V2.2) before (a) and after (b) a +120-mV/1-s depolarizing pulse in control cells (top) and in cells expressing Dr-VSP (bottom). Green lines (b’) are b current traces scaled to the peak amplitude of a current. Bottom, summary of the time constants for current inactivation (n = 5, *P < 0.01, compared to current a).

(B) Elevated PIP₂ levels slow Ca_V2.2 channel inactivation. Ca_V2.2 currents during 500-ms test pulses to +10 mV or +30 mV were measured in control cells with no Dr-VSP and in cells transfected with PIPKIγ. Typical current traces for each test voltage are overlaid.

(C) Summary of inactivation time constants (τ) for Ca_V2.2 current during the +10-mV and +30-mV test pulses with different PIP₂ levels in the plasma membrane. Data are mean ± SEM (n = 5 for control, n = 7 for PIPKIγ at both test pulses).

(D) Elevated PIP₂ levels slow Ca_V2.2 channel activation. Activation of Ca_V2.2 channels during the depolarization to +10 mV was measured in control cells with no Dr-VSP and in cells co-transfected with PIPKIγ. Currents showing similar amplitude are superimposed.

(E) Current–voltage (I-V) relations of Ca_V2.2 channels in control (open circle) and PIPKIγ-transfected (closed circle) cells. Relative currents are plotted against test potential. Points are mean ± SEM (n = 8 for control; n = 9 for PIPKIγ). See also Figure S5.

DISCUSSION

Using direct enzymatic methods to modify PIP₂ levels quickly in living cells, we have developed compelling support for the hypothesis that the endogenous PIP₂ of a cell maintains a high activity of Ca_V1.2, Ca_V1.3, Ca_V2.1, and Ca_V2.2 channels and that physiological reductions of PIP₂ immediately decrease the channel activity: (1) Irreversible depletion of endogenous membrane PIP₂ using iRap-induced translocation of INP54p 5-phosphatase to the plasma membrane irreversibly decreased the whole-cell Ca_V currents. (2) Reversible depletion of membrane PIP₂ by the activation of Dr-VSP reversibly decreased Ca_V currents in less than 1 s, with little change in voltage-dependent channel gating. (3) Elevating levels of membrane PIP₂ by transfecting with PIP 5-kinase significantly blunted the channel inhibition by Dr-VSP and accelerated recovery from inhibition ~4-fold. (4) Attenuating the endogenous PIP 5-kinase activity using nonhydrolyzable ATP analogs had parallel inhibitory effects on measured PIP₂ resynthesis and on recovery of channels from inhibition. And (5) the Dr-VSP-mediated PIP₂ depletion and the channel inhibition, as well as subsequent PIP₂ resynthesis and channel recovery, developed with similar time courses in single cells. This correspondence was especially tight for Ca_V1.3 channels. The loss of Ca_V1.3 current tracks the loss of PIP₂ within milliseconds. Together these new observations show that the activities of several HVA Ca_V channels depend on endogenous membrane PIP₂ in intact cells. This is the first direct evidence that L-type channels participate in such regulation. As a caveat, we note that we are reporting tests with the β3 and α2δ1 accessory subunits and specific splice variants of the α1 subunits. Since several forms of channel modulation are known to be affected by the subtypes of each subunit, the quantitative conclusions here can be applied strictly only to the subunits we actually tested (e.g., Raingo et al., 2007).

Membrane PIP₂ is a Modulatory Cofactor for Ca_V Channel Activity

The central question that motivated our study is whether a downstream signal of PLCβ, PIP₂ depletion, is important for signaling to Ca_V channels. We have studied the most prominent Ca_V channel types of native rat SCG neurons α1B/β3/α2δ1 (Ca_V2.2e[37b]) and α1D/β3/α2δ1 (Ca_V1.3e) (Lin et al., 1996; Bell et al., 2004) in reconstituted systems. Our data show how membrane PIP₂ turnover modulates these HVA Ca_V channels in living cell membranes and reveal similarities and novel differences between various subtypes of Ca_V channels in the modulation by PIP₂. The activity of these channels is significantly decreased by conversion of PIP₂ to PIP and remains inhibited until the PIP₂ is resynthesized from PIP by endogenous PIP 5-kinases. Thus the anionic phosphoinositide PIP₂ is a cofactor required for full channel activity. Our data suggest that the channels must be in equilibrium with plasma membrane pools of PIP₂ on a time scale much shorter than the 100 ms that it takes for Dr-VSP to depress their currents. There must be a protein-lipid binding interaction of low affinity. The short time intervening between PIP₂ dephosphorylation and the channel response argues against indirect actions such as downregulation of channels by endocytosis. The magnitude of inhibition is greater via M₁ receptors than with exogenous 5-phosphatase-dependent PIP₂ depletion.

Our experiments with intact cells did not duplicate all the phenomena reported for excised patches with Ca_V2.1 and Ca_V2.2 channels (Wu et al., 2002; Gamper et al., 2004). In the excised patch experiments, tail currents ran down nearly 100% in 2 min, and almost all of the tail current could be restored by direct addition of PIP₂. Further, the addition of PIP₂ induced a rightward shift by ~40 mV in the voltage dependence of channel activation. Finally the rightward shift was blocked in conditions favoring phosphorylation by cAMP-dependent protein kinase. In the intact-cell experiments of Gamper et al. (2004) and in our experiments, depletion of PIP₂ suppressed current only partially and any rightward shift with excess PIP₂ was <10 mV. Perhaps in whole-cell experiments, some channel phosphorylations are preserved, or possibly other cytoplasmic factors make the channels less drastically PIP₂ sensitive. Perhaps also continuing PIP₂ synthesis prevents full depletion of the PIP₂. Indeed, when we did experiments with non-hydrolyzable ATP analogues, the inhibition by VSP tended to be larger and cumulative.

Activating Dr-VSP removes the 5 phosphate from PIP₂ and produces a transient rise of membrane PI(4)P that then decays as it is converted back into PIP₂ (Halaszovich et al., 2009). Our finding that channel currents fall during the transient PIP₂ depletion means that PI(4)P is not as effective as PIP₂ in supporting channel activity. Since currents are inhibited by only 33-60% for the three channels studied, it is still possible that the elevated PI(4)P or other acidic phospholipids also support channel activity but significantly less well than resting levels of PIP₂. Direct experiments with other enzymes would be needed to test that hypothesis.

In our experiments, N-type channels were inhibited more than L-type channels by PIP₂ depletion, both by iRap-induced translocation and by Dr-VSP. Unexpectedly, the maximum inhibition by Dr-VSP was ~25-35% larger than the inhibition by iRap-induced phosphatase translocation. We suggest that the difference arises from a small basal PIP₂ 5-phosphatase activity of the translocatable INP54p that partially depletes membrane PIP₂ already before rapamycin is added. According to our kinetic models (Suh et al, 2004), even a 1% resting activity at the membrane would lower the PIP₂ level by 20%, enough to reduce the channel currents partially. This would make the subsequent iRap-induced inhibition smaller. Thus, we estimate using the Dr-VSP results that about 55% of Ca_V2.2 and 35% of Ca_V1.2 and Ca_V1.3 current is lost when endogenous PIP₂ is depleted. By comparison, about 75% of Ca_V2.2 and 55-65% of Ca_V1.2 and Ca_V1.3 current is lost when M₁ muscarinic receptors are activated. We propose that a significant fraction but not the entire muscarinic inhibition is due to PIP₂ depletion. Previously, we and others showed that activation of M₁ or M₃ muscarinic receptors significantly depletes membrane PIP₂ (Willars et al., 1998; Horowitz et al., 2005; Winks et al. 2005; Jensen et al., 2009). The depletion is >>90% as assayed by translocation or loss of FRET from PH-domain probes and by direct biochemical methods. In this multiple-pathway theory of muscarinic inhibition, there would also be several components to the post-agonist recovery as each of the underlying messenger systems make its own recovery. Commonly discussed additional candidate messengers that do act on Ca_V1 or Ca_V2 family channels are divalent ions, Gβγ subunits, arachidonic acid, and protein kinase C (Delmas et al., 2005; Michailidis et al., 2007; Roberts-Crowley et al., 2009). The alternative hypothesis, which we regard as unlikely on kinetic grounds, is that stronger inhibition by PLC simply reflects an inability of our phosphatases tools to deplete PIP₂ as much as PLC does.

Mechanisms for the PIP₂ actions on Ca_V channels

What is PIP₂ loss doing to channels to decrease the net current they carry? We considered two possibilities with negative results. We considered whether PIP₂ loss makes channels more susceptible to modulation by Gβγ subunits. This possibility seems unlikely both because the PIP₂ effect was unchanged in mutant channels where the Gβγ binding site was crippled and because the inhibition by PIP₂ depletion was not relieved by large positive “facilitating” pulses the way Gβγ inhibition would be. We also considered whether PIP₂ loss enhances VDI enough to account for the suppression of Ca²⁺ currents. Although PIP₂ depletion did speed development of VDI, and PIP₂ augmentation slowed it, the component of inhibition due to PIP₂ depletion was not reversed by large hyperpolarizing conditioning pulses that were sufficient to remove normal VDI.

Our kinetic results with Dr-VSP suggest that Ca_V1.3 channel activity follows changes in the membrane PIP₂ level closely in a simple linear manner. They can be described by a model with low-affinity, rapid, first order, non-cooperative binding of PIP₂ to Ca_V1.3 channels, where each bound PIP₂ contributes a certain increment to the channel activity. Our data would be consistent with a model having only one facilitatory PIP₂ site on Ca_V1.3 channels, but there also could be several. The Ca_V2.2 channels behave differently. Their activity possibly falls faster than PIP₂ is depleted and certainly recovers much slower than PIP₂ is regenerated. Such behavior could reflect a combination of a need for more than one bound PIP₂ for activity, slow rebinding of PIP₂ to channels, or the involvement of other PIP₂-sensitive messenger signals. Working with Ca_V2.1 channels, Wu et al. (2002) considered a model with two PIP₂ binding sites, one with facilitatory and the other with inhibitory effects. Our experiments were done very differently from theirs, and our data are insufficient to discuss such details but we did not encounter any compelling evidence for inhibitory actions of PIP₂. The structure, number, and influences of PIP₂ binding sites on any ion channels are questions for future work, but as a working hypothesis we would consider a model with at least two facilitatory sites on the Ca_V2.2 channel complex both of which need to be occupied to see enhancement by PIP₂. That would make cooperative kinetics in which channel activity falls faster than PH domain FRET ratio during inhibition and rises more slowly than FRET ratio during recovery.

Conclusions

Slow modulation of Ca²⁺ channels by M₁ muscarinic receptors and more generally by any G_q-coupled receptor uses multiple signaling pathways. We employed a strategy that keeps the cell intact yet is able to vary membrane PIP₂ quickly with minimum production of other distracting messages, especially with a novel use of a voltage-dependent phosphatase. We compensated for effects of voltage-dependent inactivation of channels on the test current amplitudes. Depending on the channel subtype, such focused experiments demonstrate that 35-55% of the Ca²⁺ channel activity is supported by PIP₂ as a cofactor. The channels do not fail with acutely reduced PIP₂ but they definitely generate larger currents when PIP₂ is at its normal endogenous level. Activation of M₁Rs removes more current than just the PIP₂-dependent component, but the PIP₂-dependent component accounts for more than half of the muscarinic modulation. Our early proposal (Bernheim et al., 1991; Mathie et al., 1992; Hille, 1994) that slow modulation of KCNQ channels and L- and N-type channels in sympathetic ganglion cells share a common pathway is partly borne out. They do use a common pathway, but the Ca²⁺-channel modulation also uses additional signals. In sum, the PIP₂ hypothesis has now been proven for four voltage-gated Ca²⁺ channel subtypes that are modulated by M₁ muscarinic receptors, and it has been shown not to apply to four other Ca_V subtypes that are not modulated by M₁ muscarinic receptors.

EXPERIMENTAL PROCEDURES

Cell Culture and Transfection

TsA201 cells (large-T-antigen transformed HEK 293 cells) were maintained in DMEM supplemented with 10% FBS and 0.2% penicillin/ streptomycin and transiently transfected using Lipofectamine 2000 (Invitrogen) with various cDNAs (See Supplementary Experimental procedures). For Ca²⁺ channel expression, cells were transfected with the α1 subunit of Ca_V, β3, and α2δ1 in a 1:1:1 molar ratio. When needed, 0.1 μg of cDNA encoding green fluorescent protein (GFP) or tetrameric red FP (DsRed) was co-transfected with the cDNA as a marker for successfully transfected cells. The next day, the cells were plated onto poly-L-lysine-coated coverslip chips, and fluorescent cells were studied within 1 - 2 days in FRET and electrophysiological experiments. Cultured SCG neurons were prepared as described (Mochida et al., 2003). Briefly, ganglia were dissected from 7-day postnatal rats, desheathed, and incubated with collagenase (0.65 mg/ml; Worthington Biochemical) in L-15 medium (Gibco) at 37°C for 40min. Following enzyme treatment, ganglia were triturated gently through a small-pore glass pipette, washed twice by low speed centrifugation, and resuspended in DMEM supplemented with 10% fetal calf serum (Gibco), 5% horse serum (Gibco), 1% penicillin–streptomycin solution (Gibco), and 25 ng/ml nerve growth factor (2.5 S; Alomone Labs Ltd.). Cells were plated on glass coverslips coated with poly-D-lysine in 35mm dish incubated at 37°C (5% CO₂). cDNA encoding Dr-VSP was microinjected into the nuclei of SCG neurons through glass micropipettes one week after plating. Successful injection was monitored by 5% fast green dye in the nucleus. N-type currents were recorded two days after injection of cDNA.

Current Recording

The whole-cell configuration of the patch-clamp technique was used to voltage-clamp and dialyze cells at room temperature (22 - 25°C). Electrodes pulled from glass micropipette capillaries (Sutter Instrument, Novato, CA) had resistances of 1.3 - 2.5 MΩ. The whole-cell access resistance was 2 - 5 MΩ, and series-resistance errors were compensated > 60%. Fast and slow capacitance was compensated prior to the applied test-pulse sequences. Ba²⁺ currents were recorded by holding the cell at −70 mV or −80 mV and applying 10-ms (or 500-ms in Figure 9) test pulses to −10 mV or +10 mV to measure Ca_V1.3 and Ca_V2.2 or Ca_V1.2 currents, respectively. Note that tsA cells do not have endogenous voltage-gated Ca²⁺ channels and all the inward Ba²⁺ current was completely blocked by application of 30 μM Cd²⁺. In the experiments with pipette solutions containing the ATP analogues AMP-PCP or AMP-PNP, we waited longer than 3 min before activating VSP proteins to allow time for the dialysis of the analogues into the cytoplasm.

Solutions and Materials

The external Ringer’s solution used for Ba²⁺ current recording and photometry contained (in mM): 150 NaCl, 10 BaCl₂, 1 MgCl₂, 10 HEPES, and 8 glucose, adjusted to pH 7.4 with NaOH. The pipette solution contained (in mM): 175 CsCl, 5 MgCl₂, 5 HEPES, 0.1 1,2-bis(2-aminophenoxy)ethane N,N,N’,N’-tetraacetic acid (BAPTA), 3 Na₂ATP, and 0.1 Na₃GTP, titrated to pH 7.4 with CsOH. For current measurements through Ca²⁺ channels in SCG neurons, the bath solution contained (in mM) 162.5 tetraethylammonium (TEA) chloride, 5 BaCl₂, 10 HEPES, 8 glucose, 1 MgCl₂, 0.0001 TTX, and 0.005 nimodipine, pH adjusted to 7.4 with TEAOH. Variations on the solutions are noted in text. Reagents were obtained as follows: oxotremorine methiodide (Oxo-M) (Research Biochemicals, Natick, MA); BAPTA (Molecular Probes, Eugene, OR); DMEM, fetal bovine serum, lipofectamine 2000, and penicillin/streptomycin (Invitrogen, Carlsbad, CA); ATP, GTP, AMP-PCP, AMP-PNP, and other chemicals (Sigma, St. Louis, MO).

Epifluorescence Photometry

Fluorescence resonance energy transfer (FRET) between CFP and YFP was measured in single cells using an epifluorescence microscope equipped with two photomultipliers in photon-counting mode as described previously (Jensen et al., 2009). Cells were studied on the inverted microscope using a 40x, 1.3 numerical aperture oil-immersion objective. Excitation arc light passed through a 0.2 ND filter and a cube with a 440 ± 10 nm excitation filter and a 465 nm dichroic mirror. The total emitted light from the entire cell image was pooled and counted after deflection to the photomultiplier tubes by two cubes in series: a 505 nm dichroic mirror with a 480 ± 15 nm filter (“short-wavelength channel”), and a 570 nm dichroic mirror with a 535 ± 12.5 nm bandpass filter (“long-wavelength channel”). For sampling, the illumination shutter was opened for 24 ms every 500 ms. The fluorescence ratio was taken as the ratio of long-wavelength to short-wavelength emission (YFP_C/CFP_C) during 440 nm illumination after corrections for background fluorescence and bleedthrough determined in separate experiments on cells transfected with single fluorophores. The subscript C is a reminder that the 440-nm excitation light is exciting CFP in both cases.

Data Analysis

Data acquisition and analysis used Pulse/Pulse Fit 8.11 software in combination with an EPC-9 patch clamp amplifier (HEKA, Lambrecht, Germany). Further data processing was performed with Excel (Microsoft, Bellevue, WA) and Igor Pro (WaveMetrics, Lake Oswego, OR). Time constants were measured by exponential fits. All quantitative data are expressed as the mean ± SEM. Comparison between two groups was analyzed using Student’s t-test, and differences were considered significant at a level P < 0.05.