Molecular basis of the PIP₂-dependent regulation of Ca_V2.2 channel and its modulation by Ca_V β subunits

We have previously reported that subcellular localization of the Ca_V β subunit plays an important role in determining the inactivation kinetics and PIP₂ sensitivity of Ca_V2.2 channels (Keum et al., 2014; Kim et al., 2015a, Kim et al., 2016). By manipulating the β2 constructs, we further examined how Ca_V β subunits determine the gating properties of the Ca_V2.2 channel depending on their subcellular localization. First, we used a palmitoylation-resistant cytosolic mutant form of β2a, β2a(C3,4S), where two palmitoylation sites (C3 and C4) in the N-terminus of the β2a subunit were mutated to serine residues (Chien et al., 1996; Hurley et al., 2000; Olcese et al., 1994; Qin et al., 1998; Figure 1A and Figure 1—figure supplement 1A). Additionally, we constructed two more membrane-recruited β2c analogs by adding membrane-targeting Lyn₁₁ (N-terminal G2-myristoylation and C3-palmitoylation modification sequence from Lyn kinase; Resh, 1994) or Lyn₁₁ plus a flexible 48 amino acid linker (Lyn-48aa) to the N-terminus of β2c. When these Ca_V β constructs were expressed in cells without the pore-forming α1B, β2a(C3,4S) was distributed through the cytosol similar to β2c. By contrast, the engineered Lyn-β2c and Lyn-48aa-β2C were localized at the PM like the membrane-anchored β2a subunit (Figure 1B, C). However, in the presence of α1B and α2δ1, all the β2 constructs were mainly distributed at the PM, probably via binding to α1B subunits (Figure 1—figure supplement 1B, C). This suggested that amino acid mutation or chimeric modification of the β2 subunit does not affect the formation of the Ca_V2.2 channel multicomplex. Next, we tested the effects of the β2 constructs on current inactivation and PIP₂ sensitivity of the Ca_V2.2 channels. PIP₂ regulation of Ca_V2.2 channel gating was measured as the difference before and after a + 120 mV depolarizing pulse using Dr-VSP (see Figure 1—figure supplement 2A). Coexpression of β2a(C3,4S) accelerated current inactivation and increased the PIP₂ sensitivity of Ca_V2.2 channels, such as those with the cytosolic β2c subunit. Expression of the chimeric Lyn-β2c slowed down the inactivation rate and decreased PIP₂ sensitivity, like the channels with the PM-anchored β2a subunit (Figure 1D, E). Interestingly, cells co-transfected with the PM-tethered chimeric Lyn-48aa-β2c showed faster current inactivation and higher PIP₂ sensitivity in Ca_V2.2 channels, which were similar to the responses of channels with the cytosolic β2c subunit. In control experiments without Dr-VSP, we confirmed that the current amplitudes of Ca_V2.2 channels with the developed β2 constructs were not significantly different before and after the depolarizing pulse (Figure 1—figure supplement 2A, B). Additionally, we verified that the effects of Dr-VSP were not due to relieving the Gβγ-mediated tonic inhibition from the Ca_V2.2 channels. As shown in Figure 1—figure supplement 2C, prepulse depolarization did not change the current amplitudes in cells intracellularly perfused with 1 mM of the G protein inhibitor GDP-β-S instead of GTP in the absence of Dr-VSP. Moreover, the Ca_V2.2 channels with GDP-β-S showed very similar PI(4,5)P₂ sensitivities to those in experiments with GTP in cells expressing Dr-VSP (Figure 1—figure supplement 2D). This suggested that 0.1 mM GTP concentration in the pipette solution was not sufficient to trigger spontaneous G protein activation or suppress Ca_V2.2 channels through Gβγ binding.

Figure 1 with 3 supplements see all

Download asset Open asset

Figure 1—source data 1

Current inactivation (r₁₀₀) and current inhibition (%) by PIP₂ depletion in N-type Ca_V2.2 channels with different subtypes of the β2 subunit.

https://cdn.elifesciences.org/articles/69500/elife-69500-fig1-data1-v1.xlsx

Download elife-69500-fig1-data1-v1.xlsx

We further examined the effects of the length of the flexible linker between Lyn and the β2c subunit on the inactivation kinetics and PIP₂ sensitivity of Ca_V2.2 channels. As shown in Figure 1—figure supplement 3, when the inserted linkers were longer than 24 aa, current inactivation was faster and current inhibition by PIP₂ depletion was stronger. Together, these data suggest that the N-terminal length of the PM-tethering Ca_V β subunit is critical in determining the inactivation kinetics and PIP₂ sensitivity of Ca_V2.2 channels.

It has been previously reported that disruption of the SH3–GK interaction in the membrane-anchored β2a subunit accelerates the channel inactivation of Ca_V2.1 channels (Chen et al., 2009). The GK domain of the Ca_V β subunit interacts directly with the AID domain in the I–II loop of Ca_V α1 subunits (Buraei and Yang, 2010; Buraei and Yang, 2013; Chen et al., 2004; Opatowsky et al., 2004; Van Petegem et al., 2004); therefore, disruption of the SH3–GK interaction in the Ca_V β subunit may increase the length between the N-terminus and the GK–AID complex through the flexible HOOK region. To test the possible effects of increased N-terminal length from the AID–GK complex on Ca_V channel gating, we constructed mutant β2a subunits in which the SH3–GK intramolecular interaction was disrupted by mutating seven amino acids in the SH3 and GK domains to alanine residues (Figure 2A). Additionally, the N-terminus was deleted to abolish membrane targeting of the β2a subunit by itself, and Lyn₁₁ was inserted to the N-terminus for membrane recruitment. Without α1B and α2δ1, both N-terminus-deleted (∆N)β2 WT and (∆N)β2 Mut, in which the SH3–GK interaction was disrupted, were expressed in the cytosol. Conversely, Lyn-(∆N)β2 WT and Lyn-(∆N)β2 Mut constructs were localized to the PM (Figure 2A, inset images). In Ca_V2.2 channels with the N-terminus-deleted mutant (∆N)β2 WT, the current exhibited fast inactivation and high PIP₂ sensitivity (Figure 2B–D). These phenomena similarly appeared in channels with the (∆N)β2 Mut. In contrast, Ca_V2.2 channels with Lyn-(∆N)β2 WT exhibited slow inactivation and weak PIP₂ sensitivity. However, the channels with Lyn-(∆N)β2 Mut exhibited fast inactivation and strong PIP₂ sensitivity, like channels with cytosolic (∆N)β2 WT and (∆N)β2 Mut (Figure 2B–D). We also confirmed that disruption of the SH3–GK interaction did not shift the current–voltage (I–V) curve of Ca_V2.2 currents (Figure 2—figure supplement 1). These data suggested that the length from the N-terminal lipid anchor to the GK domain of β subunit is crucial in determining the inactivation rate and PIP₂ sensitivity of Ca_V2.2 channels.

Figure 2 with 1 supplement see all

Download asset Open asset

Figure 2—source data 1

Current inactivation (r₁₀₀) and current inhibition (%) by PIP₂ depletion in N-type Ca_V2.2 channels with the engineered β2 construct.

https://cdn.elifesciences.org/articles/69500/elife-69500-fig2-data1-v1.xlsx

Download elife-69500-fig2-data1-v1.xlsx

To further examine the functional role of length between lipid anchor and GK domain on Ca_V channel gating in live cells, we developed new chimeric β2 constructs by applying the rapamycin-induced dimerizing system FK506-binding protein (FKBP) and FKBP–rapamycin-binding (FRB) protein (Banaszynski et al., 2005; Inoue et al., 2005; Suh et al., 2006). As shown in Figure 3A, FKBP and FRB proteins irreversibly assembled to form a ternary complex upon application of rapamycin, which led to shortening of the length between the lipid anchor Lyn₁₁ and GK–AID domains. We fused a Förster resonance energy transfer (FRET) probe YFP to the C-terminus of all β2 chimera to investigate whether the FKBP domain was really translocated to the PM to make a Lyn₁₁-FRB and FKBP complex closely after rapamycin addition (Figure 3A, right diagram). In experiments with the β chimera without the FKBP domain (Control: Lyn-FRB-HOOK-GK), both FRETr and the current amplitude of Ca_V2.2 channels were not changed by rapamycin addition (Figure 3B). Consistently, rapamycin treatment did not affect current inactivation and the PIP₂ sensitivity of Ca_V2.2 channels in these cells (Figure 3C–F). In contrast, in Ca_V2.2 channels with Lyn-FRB-HOOK-GK-FKBP (RF), rapamycin treatment irreversibly enhanced the FRETr signal and increased the current amplitude of Ca_V2.2 channels (Figure 3B, middle and Figure 3—figure supplement 1). Moreover, rapamycin treatment reduced the current inactivation and PIP₂ sensitivity of Ca_V2.2 channels (Figure 3C–F). However, in Ca_V2.2 channels with Lyn-FRB-HOOK-GK-Linker-FKBP (RCF), where a 194-aa linker was inserted between GK and FKBP, rapamycin enhanced the FRETr signal without causing significant changes in the current amplitude (Figure 3B, right and Figure 3—figure supplement 1). The effects of rapamycin on inactivation kinetics and PIP₂ sensitivity were much weaker in Ca_V2.2 channels with RCF when compared with those in channels with RF (Figure 3C–F). This suggested that rapamycin-induced dimerization may be insufficient to shorten the length between the lipid anchor and isolated GK domain of β subunit in channels with RCF.

Figure 3 with 1 supplement see all

Download asset Open asset

Figure 3—source data 1

Time courses of Ca_V2.2 currents and Förster resonance energy transfer (FRET) ratio.

https://cdn.elifesciences.org/articles/69500/elife-69500-fig3-data1-v1.xlsx

Download elife-69500-fig3-data1-v1.xlsx

Figure 3—source data 2

Current inactivation (r₁₀₀) and current inhibition (%) by PIP₂ depletion in Ca_V2.2 channels with rapamycin-induced translocatable β2 chimeric constructs before and after rapamycin.

https://cdn.elifesciences.org/articles/69500/elife-69500-fig3-data2-v1.xlsx

Download elife-69500-fig3-data2-v1.xlsx

Next, we measured the effects of the N-terminal length of PM-tethered β subunit on Ca_V2.2 channel activity by inserting flexible linkers of various lengths between Lyn₁₁ and the GK domain of β2 (Figure 4A). The inserted linkers were unstructured flexible peptides (see Figure 4—figure supplement 1); therefore, the length of the linkers was calculated using the worm-like chain (WLC) model (see Methods). Our results showed that both the current inactivation and PIP₂ sensitivity of Ca_V2.2 channels became gradually stronger as the inserted flexible linkers became longer (Figure 4B–D). Consistently, the current activation was gradually accelerated by the increase in linker length (Figure 4—figure supplement 2). However, no additional difference was detected in channels with the membrane-tethered Lyn-43aa-GK subunit when compared with the cytosolic GK subunit. This indicated that the GK domain with the length of the inserted 43-aa linker is sufficient to act like the cytosolic Ca_V β subunit (Figure 4B–D). Interestingly, the PIP₂ sensitivity and inactivation kinetics of Ca_V channels were differentially regulated by the length between the lipid anchor and the GK domain: the channels with Lyn-43aa-GK showed faster inactivation than the channels with Lyn-22aa-GK, whereas the PIP₂ sensitivity of the two channels was not significantly different (Figure 4B–E). Additionally, our data analysis indicated that the biophysical gating properties of Ca_V2.2 channels with a membrane-anchored β2a subunit were similar to those of channels with Lyn-9aa-GK. Furthermore, the gating properties of Ca_V2.2 channels coupled with cytosolic β2c were similar to those of channels with Lyn-20aa-GK (Figure 4E).

Figure 4 with 4 supplements see all

Download asset Open asset

Figure 4—source data 1

Current inactivation (r₁₀₀) and current inhibition (%) by PIP₂ depletion in Ca_V2.2 channels with the engineered β2 GK derivatives.

https://cdn.elifesciences.org/articles/69500/elife-69500-fig4-data1-v1.xlsx

Download elife-69500-fig4-data1-v1.xlsx

Previous studies have reported that subcellular localization of the Ca_V β subunit is important in determining the current density of Ca_V channels, where Ca_V channels with the membrane-anchored β subunit show relatively higher current density than channels with the cytosolic β subunit (Suh et al., 2012). In line with this, we found that the current density of Ca_V2.2 channels with β2a was significantly higher than that of channels with β2c (Figure 4—figure supplement 3A–C). Therefore, we tested whether the current density of Ca_V channels was dependent on the N-terminal length. Ca_V2.2 channels showed slightly decreased current density that was dependent on the expansion of the flexible linker length between Lyn and the GK domain alone (Figure 4—figure supplement 4A, C). This phenomenon was observed in channels with the whole β2c subunit with Lyn (Figure 4—figure supplement 3D–F). We tested whether the length between N-terminal lipid anchor and GK domain affected the voltage-dependent gating of Ca_V channels. Voltage-dependent activation of Ca_V2.2 channels with Lyn-linker-GK derivatives showed a greater shift to positive voltage as the inserted flexible linkers increased in length (Figure 4—figure supplement 4B, D). This suggested that incremental increases in linker length lead to a decreased voltage sensitivity. There was no difference in the current density and voltage-dependent activation between Ca_V channels with the Lyn-43aa-GK and GK subunit. Together, these results suggested that differential regulation of Ca_V2.2 channel gating by β subunits is mainly determined by the anchoring properties of the β subunits to PM.

How does the N-terminal length of the PM-tethering Ca_V β subunit regulate Ca_V channel gating? Recently, Kaur et al., 2015 have reported that a polybasic motif consisting of four basic amino acids within the C-terminal end of the I–II loop of L-type Ca_V1.2 channels interacts with membrane phospholipids, including PIP₂. Additionally, the putative PIP₂-binding site is conserved in the I–II loop of N-type Ca_V2.2 channels (Figure 5—figure supplement 1). We examined whether the polybasic motif affects the PIP₂ sensitivity of Ca_V2.2 channels. First, we eliminated the potential phospholipid-binding motif from the Ca_V2.2 channel I–II loop by mutating the four polybasic residues to alanine (4A α1B) (Figure 5A). In Ca_V2.2 channels with the β2a subunit, the inactivation kinetics of the current did not differ between WT α1B and 4A α1B (Figure 5B, C, left). However, in Ca_V2.2 with β2c, the inactivation rate was slower in 4A α1B channels (Figure 5B, C, right). The effects of PIP₂ depletion on current amplitude were also measured in these channels. In control experiments without Dr-VSP, the current of WT or 4A-mutant Ca_V2.2 channels did not significantly differ before and after a + 120-mV depolarizing pulse in cells with either β2a or β2c subunits (Figure 5D). By contrast, PIP₂ depletion by Dr-VSP activation similarly inhibited the Ca_V current by approximately 5% in cells expressing either WT or 4A Ca_V2.2 channels with a PM-anchored β2a subunit (Figure 5E, left). This indicated the presence of another PIP₂-binding site in the α1B subunit other than this polybasic motif in I–II loop. On the other hand, the PIP₂ sensitivity in channels with β2c was dramatically reduced in 4A channels, indicating that the polybasic motif in the I–II loop plays a key role in PIP₂ regulation of Ca_V2.2 channels with the cytosolic β subunit (Figure 5D, E). However, in cells expressing 4A Ca_V2.2 channels with β2c, we observed another ~5% current inhibition by PIP₂ depletion. This was similar to the Ca_V2.2 channels with β2a.

Figure 5 with 3 supplements see all

Download asset Open asset

Figure 5—source data 1

Current inactivation (r₁₀₀) and current inhibition (%) by PIP₂ depletion in cells expressing α1B WT and 4A mutants with β2a and β2c.

https://cdn.elifesciences.org/articles/69500/elife-69500-fig5-data1-v1.xlsx

Download elife-69500-fig5-data1-v1.xlsx

Next, we investigated whether the polybasic motif affects the PIP₂ sensitivity of Ca_V2.2 channels with Lyn-β2c and Lyn-48aa-β2c (Figure 5—figure supplement 2). Similar to β2a, we did not detect any significant differences in current inactivation and PIP₂ sensitivity between WT and 4A mutant Ca_V2.2 with Lyn-β2c (Figure 5—figure supplement 2). Conversely, WT Ca_V2.2 channels with Lyn-48aa-β2c exhibited faster inactivation and higher PIP₂ sensitivity, which was similar to the responses of Ca_V2.2 channels with cytosolic β2c. However, in cells expressing 4A mutant Ca_V2.2 channels with Lyn-48aa-β2c, the current inactivation was slowed and the PIP₂ sensitivity was decreased to ~5% (Figure 5—figure supplement 2). The PIP₂ sensitivities of 4A Ca_V2.2 channels with Lyn-β2c and Lyn-48aa-β2c did not significantly differ and were similar to that of WT channels with Lyn-β2c. Consistent with the data in Figure 5, these results suggested that the polybasic motif within the I–II loop interacts with membrane PIP₂ in Ca_V2.2 channels with β2c-like Lyn-48aa-β2c, but not with β2a-like Lyn-β2c subunits. On the other hand, in channels with the β2a subunit, there was no significant difference in the voltage-dependent activation between WT α1B and 4A α1B (Figure 5F, G). However, the activation of 4A α1B with the β2c subunit was significantly shifted toward the hyperpolarization direction when compared with WT α1B channels with β2c (Figure 5F, G). In addition, the activation curve of 4A α1B with β2c was similar to the curves of WT and 4A α1B with β2a (Figure 5F, G). Together, our data suggested that two different PIP₂-interacting sites with differential PIP₂ sensitivities exist in Ca_V2.2 channels. More importantly, our data indicate that PIP₂ interacts with the polybasic motif when Ca_V2.2 is expressed with cytosolic β subunits but not when expressed with lipidated membrane-anchored β subunit.

Finally, we determined whether other arginine residues in the distal region of polybasic motif also affected the PIP₂ sensitivity of Ca_V2.2 channels (Figure 5—figure supplement 3). For this, two arginine residues (R476 and R477) near the polybasic motif were replaced with alanine (α1B R476,477A) (Figure 5—figure supplement 3A). We also constructed a α1B R465,466A by mutating only two arginine residues (R465 and R466) in the polybasic motif (R465, R466, K469, and R472) (Figure 5—figure supplement 3A). In Ca_V2.2 channels with the β2a subunit, we did not detect any significant differences in current inactivation and PIP₂ sensitivity among WT α1B, α1B R465,466A, and α1B R476,477A (Figure 5—figure supplement 3B–E). However, in Ca_V2.2 with β2c, the inactivation rate was slower and the PIP₂ sensitivity was weaker in both α1B R465,466A and α1B R476,477A compared to WT α1B (Figure 5—figure supplement 3B–E).

To examine whether the polybasic motif influenced the G_q-coupled modulation of Ca_V2.2 channels, we applied the muscarinic acetylcholine receptor agonist, oxotremorine-M (Oxo-M), to cells co-transfected with the M₁ muscarinic receptor (M₁R) (Figure 6). Since the M₁R stimulation suppressed Ca_V2.2 channels through both Gβγ binding to channels and PIP₂ depletion (Keum et al., 2014), we then used a Gβγ-insensitive chimeric Ca_V2.2 channel construct, α1C-1B, to examine the effect of PIP₂ depletion alone on channel regulation (Figure 6). In this chimera construct, the N-terminus of Ca_V2.2 (α1B), which contains one of the Gβγ interaction sites, is replaced by the N-terminus of Ca_V1.2 (α1C) (Agler et al., 2005). M₁R activation inhibited the current by approximately 5% in cells expressing either α1C-1B WT or 4A channels with β2a subunit, which were similar to the responses of regulation by Dr-VSP-mediated PIP₂ depletion in those channels (Figure 6B, C). However, consistent with the results for Dr-VSP-induced channel modulation, current suppression was much weaker in α1C-1B 4A channels with β2c than in α1C-1B WT with β2c (Figure 6B, C). We confirmed that the current suppression by M₁R activation were not recovered in both α1C-1B WT and α1C-1B 4A channels by a prepulse regardless of the coupled β2 isotypes (Figure 6D, E). We additionally used G_i-coupled M₂ muscarinic receptor (M₂R) to further examine whether the polybasic motif in I–II loop affects the Gβγ-mediated modulation of Ca_V2.2 channels (Figure 6—figure supplement 1). M₂R activation inhibited the currents evoked by a + 10-mV test pulse without significant difference between WT and 4A α1B with β2a or β2c (Figure 6—figure supplement 1B, C). M₂R activation commonly slowed down the activation kinetics of Ca_V2.2 currents (Figure 6—figure supplement 1D). We have previously reported that subcellular localization of the Ca_V β subunit plays important roles in determining the Gβγ-dependent inhibition of Ca_V2.2 channels; membrane-anchored β2a subunit changes Ca_V2.2 channels are more sensitive to Gβγ-mediated voltage-dependent inhibition, whereas cytosolic β2b and β3 subunit changes channels are less sensitive to Gβγ-mediated voltage-dependent inhibition (Keum et al., 2014). In Ca_V2.2 channels with the β2a subunit, the recoveries from Gβγ-mediated inhibition did not significantly differ between WT α1B and 4A α1B (Figure 6—figure supplement 1E, F). However, in Ca_V2.2 with the β2c subunit, there was less recovery from Gβγ-mediated inhibition in α1B WT than in α1B 4A (Figure 6—figure supplement 1E, F). Recovery from M₂R-mediated inhibition in 4A α1B with β2c was similar to the values of WT and 4A α1B with β2a (Figure 6—figure supplement 1E, F).

Figure 6 with 1 supplement see all

Download asset Open asset

Figure 6—source data 1

Current inhibition (%) of α1C-1B WT and 4A mutants by M₁R or Dr-VSP activation in cells expressing with β2a and β2c.

Summary of the prepulse experiments in before and Oxo-M perfused cells with α1C-1B WT and 4A mutants with β2a or β2c subunits.

https://cdn.elifesciences.org/articles/69500/elife-69500-fig6-data1-v1.xlsx

Download elife-69500-fig6-data1-v1.xlsx

Recently, the cryo-electron microscopic structure of human Ca_V2.2 complex composed of α1B, α2δ1, and β3 subunits was revealed at a resolution of 3.0 Å (Dong et al., 2021; Gao et al., 2021). These studies have shown that the 5-phosphate group of membrane PIP₂ interacts with two basic residues (R584 and K587) within S4_II domain of α1B. We examined whether the two basic residues affect the PIP₂ sensitivity of Ca_V2.2 channels. First, we constructed neutralized mutant α1B subunits in which the two basic residues in S4_II were replaced by alanine residues (α1B RA/KA) (Figure 7A). In Ca_V2.2 channels with β2a, the inactivation kinetics of the current were not changed in α1B and α1B RA/KA, regardless of the 4A mutation (Figure 7B, C). In Ca_V2.2 with β2c, WT α1B RA/KA showed faster inactivation than WT α1B, whereas 4A α1B RA/KA showed much slower but similar inactivation to those of α1B and α1B RA/KA with β2a (Figure 7B, C). Additionally, the effects of PIP₂ depletion on current amplitude were measured in these mutant channels. Mutation of the two basic residues in S4_II completely abolished the Dr-VSP-mediated current inhibition in cells expressing WT α1B RA/KA or 4A α1B RA/KA with the β2a subunit, while there was approximately 5% inhibition in cells expressing WT and 4A α1B with β2a (Figure 7D, E). Importantly, PIP₂ depletion significantly inhibited the currents in cells expressing WT α1B RA/KA and β2c, whereas 4A α1B RA/KA exhibited no current inhibition, like α1B RA/KA with β2a (Figure 7D, E). Since the mutation of two basic residues changes the gating charges of S4_II voltage-sensor domain, we additionally tested if other charge residues within the S4_II similarly affects PIP₂ sensitivity of Ca_V2.2 channels. We eliminated two adjacent arginine residues (R578 and R581) in S4_II by replacing with alanine (α1B R578,581A) (Figure 7—figure supplement 1A). In both WT α1B and α1B R578,581A channels with β2a or β2c, there was no significant changes in the current inhibition by PIP₂ depletion (Figure 7—figure supplement 1B, C), suggesting that the R578 and R581 charge residues near the PIP₂-binding pocket were not involved in the PIP₂ interaction. Next, we examined the functional role of the PIP₂-binding site within S4_II in the voltage-dependent activation of Ca_V2.2 channels. Regardless of β2 isotype, the activation curves were significantly shifted toward the depolarization direction in both WT α1B RA/KA and 4A α1B RA/KA (Figure 7F, G). Together, our results suggest that the two basic residues within the S4_II domain consistently interact with PIP₂ regardless of the coupled β2 isotype. Additionally, PIP₂-binding to the S4_II-binding pocket is important in maintaining stable Ca_V2.2 channel gating.

Figure 7 with 1 supplement see all

Download asset Open asset

Figure 7—source data 1

Current inactivation (r₁₀₀), current inhibition (%) by PIP₂ depletion and the V_1/2 of normalized steady-state activation in cells expressing WT α1B, WT α1B RA/KA, 4A α1B, and 4A α1B RA/KA with β2a or β2c.

https://cdn.elifesciences.org/articles/69500/elife-69500-fig7-data1-v1.xlsx

Download elife-69500-fig7-data1-v1.xlsx

Previous studies have proposed a bidentate model for the PIP₂ modulation of N-type Ca_V2.2 channel regulation (Heneghan et al., 2009; Hille et al., 2015; Mitra-Ganguli et al., 2009; Roberts-Crowley and Rittenhouse, 2009). In this model, lipidation on the N-terminus of Ca_V β subunits disrupts the hydrophobic interaction between the two fatty acyl chains of PIP₂ and Ca_V2.2 channels, and thus reduces current inhibition by PIP₂ depletion. For example, β2a subunits interact with the PM through two palmitoyl fatty acyl chains in the N-terminus, leading to competition in binding to Ca_V channels with the fatty side chains of PIP₂. This competition removes PIP₂ from the channel-binding site. Thus, Ca_V channels with β2a are uncoupled from the membrane PIP₂ and show lower PIP₂ sensitivity to PIP₂ depletion. By contrast, non-lipidated β3 subunits expressed in the cytosol do not interrupt the interaction between the fatty acyl chains of PIP₂ and Ca_V2.2 channels, and show high PIP₂ sensitivity of channels (Heneghan et al., 2009; Hille et al., 2015; Suh et al., 2012). Consistently, we found that when the β3 subunits were anchored to the PM by adding the lipidation signal of Lyn to the N-terminus, the engineered Lyn-β3 construct decreased the PIP₂ sensitivity of Ca_V2.2 channels, similar to β2a. The Lyn₁₁ domain incorporates into the PM through the G2-myristoylated and C3-palmitoylated lipid anchors; therefore, Lyn-β3 mimics β2a in competing with PIP₂ for the hydrophobic Ca_V2.2 channel interaction. Conversely, the lipid anchor of Lyn-48aa-β2c may be localized far from the channel complex because of its long N-terminal flexible linker, suggesting that these mutant subunits cannot disrupt the hydrophobic interaction between PIP₂ and channels.

Our results provide advance information about the bidentate model. First, we confirmed that two distinct PIP₂-interacting sites were preserved in the Ca_V2.2 channel: the binding pocket in VSD_II and phospholipid-binding site in the I–II loop. Our data are consistent with that the 5-phosphate group of membrane PIP₂ interacts with the two basic residues within the S4_II domain of Ca_V2.2 channels regardless of β2 isotype. The additional interaction of PIP₂ with the nonspecific phospholipid biding site in the distal I–II loop of Ca_V channels was mainly observed in Ca_V2.2 channels with the cytosolic Ca_V β2c subunit. Our data indicate that PIP₂-binding to the I–II loop phospholipid-binding site is selectively disrupted by the lipid anchor of membrane-anchored β2a. The hydrophobic interaction of the palmitoyl or myristoyl groups of Ca_V β2a or Lyn-β constructs with channel complex may be the cause of PIP₂ release from the lower-affinity I–II loop phospholipid-binding site (Roberts-Crowley and Rittenhouse, 2009). When PIP₂ interacts with the VSD_II PIP₂-binding site of Ca_V2.2 channels complexed with β2a, the PIP₂ sensitivity of the channels dramatically decreased to approximately 5%. Our results suggested that this minimal PIP₂ sensitivity specifically caused by PIP₂ degradation on VSD_II-binding pocket by Dr-VSP activation.

This work suggests that the PIP₂ sensitivity of the Ca_V2.2 channel is mainly affected by the length between the lipid anchor and GK domain of the Ca_V β subunit. Although both Lyn-β2c and Lyn-48aa-β2c are localized at the PM, the PIP₂ sensitivity and inactivation kinetics of Ca_V2.2 channels are significantly different from each other: Ca_V2.2 channels with Lyn-β2c subunits exhibited relatively slower inactivation kinetics and lower PIP₂ sensitivity, similar to channels with the membrane-anchored β2a subunit. By contrast, Ca_V2.2 channels with Lyn-48aa-β2c subunits exhibited faster inactivation kinetics and higher PIP₂ sensitivity, similar to channels with the cytosolic β2c subunit. Similarly, disruption of the SH3-GK interaction in the membrane-anchored β2a subunit accelerated current inactivation and increased the current inhibition by PIP₂ depletion. Moreover, real-time translocation of the lipid anchor, Lyn₁₁, to the channel complex by rapamycin-inducible dimerization systems slowed the inactivation and decreased the PIP₂ sensitivity of Ca_V2.2 channels. Inversely, incremental increases in flexible linker length between the lipid anchor and GK domain of Ca_V β2 subunits gradually accelerated the inactivation kinetics and increased the PIP₂ sensitivity of Ca_V2.2 channels. However, the mechanism by which the physical distance from the PM lipid to GK domain of the Ca_V β subunit affects the PIP₂ sensitivity of the Ca_V2.2 channel is not fully understood yet. Another possibility is that torsional rigidity of the linker domain may be different depending on the length and thus differently restrict the cytoplasmic movement of Ca_V β subunit as well as the gating of Ca_V2.2 channels.

Colecraft et al. have reported that chemically induced anchoring of intracellular loops of the channels to the PM can modulate the gating of the HVA Ca²⁺ channel (Subramanyam and Colecraft, 2015; Yang et al., 2013). They have shown that PdBu-induced translocation to the PM of chimeric β3-C1_PKCγ, which is assembled by fusing the C1 domain of PKCγ to the C-terminus of the β3 subunit, leads to the inhibition of the Ca_V2.2 current. Conversely, the C1_PKCγ-β3 subunit, which is assembled by adding C1_PKCγ to the N-terminus of the β3 subunit, has no effect on the current (Yang et al., 2013). These studies suggest that the polarity of the PM-targeting domain may play an important role in determining the Ca_V2.2 channel gating; however, the molecular basis of the differential regulation mechanism remain unclear. On the basis of our results, we speculate that the C1_PKCγ-β3 form may be insufficient to disrupt the interaction with between phospholipid-binding site and PIP₂ in Ca_V2.2 channels because the length from the C1_PKCγ and the GK domain of the β3 subunit is 175 aa. This could be too long to interfere the interaction between PIP₂ and Ca_V2.2 channels.

Recently, Gao et al., 2021 have shown that two basic gating charge residues (R584 and K587) within the S4_II domain of human Ca_V2.2 channel interact with the 5-phosphate group of membrane PIP₂. In our present work, we found that mutation of the two residues (RA/KA) in the S4_II domain completely blocked the Dr-VSP-induced current suppression in channels with β2a and shifted the voltage-dependent activation curve toward the depolarization direction regardless of Ca_V β2 isotype. The cryo-EM structure does not show the nonspecific PIP₂-binding site in the channels probably because it is located in the flexible I–II loop. We hypothesize that the polybasic residues in the I–II loop tether to the anionic phospholipids through the electrostatic interaction and this dipole–dipole interaction may contribute to the low-affinity phospholipid-binding site (Yeon et al., 2018). In contrast, the VSD_II PIP₂-binding site forms a pocket-like structure inside the S4_II domain and covered by the AID domain in the cytosolic side (Dong et al., 2021; Gao et al., 2021), which could stabilize the domain in a high-affinity PIP₂ interacting site. Thus, it is possible that the PIP₂ molecule inside the VSD_II PIP₂-binding pocket is relatively less accessible to the degradation by phospholipase C or Dr-VSP, leading to the lower PIP₂ sensitivity in Ca_V2.2 channels.

In conclusion, our findings provide new insights on the regulatory mechanism of Ca_V2.2 channel gating by Ca_V β subunits. Our recent study has reported that when intracellular Ca²⁺ is increased by depolarizing the cells or activating G_q-coupled receptors, the high intracellular Ca²⁺ concentration induces a dissociation of the N-terminus of the Ca_V β2e subunit from the PM. This increases both the inactivation kinetics and PIP₂ sensitivity of Ca_V2.2 channels (Kim et al., 2016). The N-terminus of the β2e subunit is anchored to the PM via electrostatic interaction with the anionic phospholipids of these PM. These studies suggest that dissociation of the β2e subunit from the membrane leads to an interaction between the I–II loop phospholipid-binding site and PIP₂, which changes the gating properties of Ca_V channels in physiological conditions. The interaction of Ca_V α1B with β subunits can be dynamically exchanged by other free β isoforms in intact cells (Yeon et al., 2018); therefore, the displacement of cytosolic β subunits by membrane-tethered β subunits on Ca_V channels will abolish the interaction with between PIP₂ and the I–II loop phospholipid-binding site via lipid anchor of membrane-tethered β subunits, which alters the Ca_V channel gating properties. Further studies are needed to investigate whether the conformational shift of the I–II loop to the membrane or cytosolic face by endogenous β subunit combinations determines Ca_V channel gating in neurons and other excitable cells.

Human embryonic kidney tsA-201 cells (large T-antigen transformed HEK293 cells; RRID:CVCL_2737) were a kind gift from Dr Bertil Hille at University of Washington. The identity of this cell line has been authenticated by STR analysis and has recurrently tested negative for mycoplasma contamination using PCR (Cosmogenetech, Daejeon, South Korea). Cells were maintained in Dulbecco modified Eagle medium (Invitrogen, CA) supplemented with 10% fetal bovine serum (Invitrogen, CA) and 0.2% penicillin/streptomycin (Invitrogen, CA) in 100 mm culture dishes at 37°C with 5% CO₂. The cells were transiently transfected with Lipofectamine 2000 (Invitrogen, CA) when the confluency of the cells reached 50–70%. For assessment of Ca_V channel expression, the cells were co-transfected with α1 of Ca_V, α2δ1, and various β2 chimera constructs in a 1:1:1 molar ratio. The transfected cells were plated onto a coverslip chip coated with poly-l-lysine (0.1 mg/ml, Sigma-Aldrich, MO) 24–36 hr after transfection. Plated cells were used for electrophysiological and confocal experiment within 24 hr after plating, as described previously (Park et al., 2017).

The following plasmids were used: The calcium channel subunits α1B of rat Ca_V2.2e[37b] (GenBank Sequence accession number AF055477) and rat α2δ1 (AF286488) were from Diane Lipscombe, Brown University, Providence, RI. Chimeric α1C-1B was generously donated by David T. Yue, Johns Hopkins University, Baltimore, MD. Mouse cDNAs of β2a and β2c were generously donated by Veit Flockerzi, Saarland University, Homburg, Germany. The Dr-VSP (AB308476) was obtained from Yasushi Okamura, Osaka University, Osaka, Japan.

Cloning of β2a-GFP, β2a(C3,4S)-GFP, and β2c-GFP was performed as previously described (Park et al., 2017). For the generation of various β2 chimera constructs, we used the one-step sequence- and ligation-independent cloning (SLIC) as a time-saving and cost-effective cloning strategy (Jeong et al., 2012). First, pEGFP-N1, pEYFP-N1, and mCherry-N1 vectors (Clontech) were linearized by KpnI restriction enzyme digestion. The cDNAs encoding β2a, β2c, Lyn, FRB, or FKBP were amplified by PCR using primers with an 18-bp homologous sequence attached to each end of the linearized vector. Primers used for β2 chimera constructs are listed in Supplementary file 1. Second, the linearized vector and PCR fragments were blended and incubated at room temperature for 2.5 min with T4 DNA polymerase (NEB, The Netherlands). Third, the DNA mixture was kept on ice for 10 min, after which competent Escherichia coli cells were transformed directly. For the deletion and point mutation of GK-SH3 interaction sites of the β2 subunit and the potential PIP₂-interaction sites of α1B, first, the α1B or β2 subunits were amplified by inverse PCR using nPfu-special DNA polymerase (Enzynomics, Daejeon, South Korea). Second, the PCR product was 5′-phosphorylated by T4 polynucleotide kinase (Enzynomics, Daejeon, South Korea) and plasmid DNA was digested by Dpn I (Agilent Technologies, Santa Clara, CA). Finally, the PCR product was ligated by T4 DNA ligase (NEB, The Netherlands). The primers used for mutagenesis are listed in Supplementary file 2. All the chimera and mutant constructs were verified by DNA sequencing (Macrogen, South Korea).

The whole-cell configuration of the patch-clamp technique was used to record Ba²⁺ currents using HEKA EPC-10 patch-clamp amplifier with pulse software (HEKA Elektronik). Electrodes pulled from glass micropipette capillaries (Sutter Instrument) had resistances of 2–4 MΩ. The whole-cell access resistance was of 2–6 MΩ, and series resistance errors were compensated by 60%. For all recordings, cells were maintained at −80 mV. The external solution contained 10 mM BaCl₂, 150 mM NaCl, 1 mM MgCl₂, 10 mM HEPES, and 8 mM glucose, adjusted to pH 7.4 with NaOH and an osmolarity of 321–350 mOsm. The internal solution of the pipette consisted of 175 mM CsCl₂, 5 mM MgCl₂, 5 mM HEPES, 0.1 mM 1,2-bis(2-aminophenocy)ethane N,N,N′,N′-tetraacetic acid (BAPTA), 3 mM Na₂ATP, and 0.1 mM Na₃GTP, adjusted to pH 7.4 with CsOH and an osmolarity of 321–350 mOsm.

All imaging examinations were performed with an LSM 700 confocal microscope (Carl Zeiss AG) at room temperature (22–25℃). The external solution for confocal imaging contained 160 mM NaCl, 2.5 mM KCl, 2 mM CaCl₂, 1 mM MgCl₂, 10 mM HEPES, and 8 mM glucose, adjusted to pH 7.4 with NaOH and an osmolarity of 321–350 mOsm. For live-cell imaging, images were obtained by scanning cells with a ×40 (water) apochromatic objective lens at 1024 × 1024 pixels using digital zoom. Analysis of line scanning of fluorescence images was performed using the ‘profile’ tool in Zen 2012 lite imaging software (Carl Zeiss Microimaging). To analyze colocalization, we performed quantitative colocalization analysis using Fiji software with the Colocalization Threshold plugin to determine the Pearson’s correlation coefficient (R). Pixel intensities were presented as 2D intensity histograms with a linear regression line and as bar graphs with mean R values. All images were transferred from LSM4 to JPEG format.

FRET experiments were performed using a monochromator (Polychrome V; TILL Photonics) with a ×40, NA 0.95 dry immersion objective lens (Olympus). Regular pulses of indigo light (438 ± 12 nm) excited the fluorescent proteins. Emission was separated into short (460–500 nm) and long (520–550 nm) wavelengths by appropriate filters and then acquired by two photomultipliers. Donor and acceptor signals obtained by photometry (TILL Photonics) were transferred to the data acquisition board (PCI-6221; National Instruments). Signal acquisition and real-time calculation of the FRET ratio were conducted by a custom program. The FRET ratio was calculated as follows:

CFP_C is the CFP emission detected by the short-wavelength photomultiplier, and YFP_C is the YFP emission detected by the long-wavelength photomultiplier, as described previously (Keum et al., 2014).

The Lyn-Linker-(additional Link) structure was suggested as an unstructured structure from the IUPRed Web-server (http://iupred.elte.hu/) (Dosztányi et al., 2005) to predict disorder tendency. To calculate the distance between the GK domain and the inner surface of the PM, the WLC model was used. This model is usually used to describe the behavior of polymers that are semi-flexible: quite stiff with successive segments pointing in roughly the same direction, and with persistence length within a few orders of magnitude of the polymer length. This model is also used to describe unstructured proteins like this linker structure (Zhou, 2001). In the WLC, the mean square end-to-end distance ${\displaystyle <R^2>}$ is written as:

where P is the polymer’s characteristic persistence length and $L_0$ is the maximum length. We used P = 0.6 and $L_0$ as (N − 1)*3.8, where N is number of amino acids in the unstructured protein (Lapidus et al., 2002). We then removed three amino acids in Lyn(MGC), which is directly connected to the membrane via palmitoylation and myristoylation. The root mean square end-to-end distance ${\displaystyle \sqrt{<R^2>}}$, which can be suggested as the average distance, was calculated. ${\displaystyle \sqrt{<R^2>}}$ was 32.7 Å for six additional linkers, 36.0 Å for 11 aa, 38.4 Å for 15 aa, 42.4 Å for 22 aa, 52.5 Å for 43 aa, and 28.2 Å for no additional linker.

Patch clamp data acquisition and analysis used Pulse/Pulse Fit 8.11 software with the EPC-10 patch clamp amplifier (HEKA Elektronik). Further data processing was performed with Igor Pro 6.2 (WaveMetrics, Inc), Excel office 365 (Microsoft), and GraphPad Prism 7.0 (GraphPad Software, Inc). All quantitative data were presented as mean ± standard error of the mean values. Comparisons between groups were analyzed by Student’s two-tailed unpaired t-test. Comparisons among more than two groups were analyzed using one-way analysis of variance (ANOVA) followed by Tukey post hoc test. Comparisons among more than two groups with two independent variables were analyzed using two-way ANOVA followed by Sidak post hoc test. Differences were considered significant at the *p < 0.05, **p < 0.01, and ***p < 0.001, as appropriate.

The funders had no role in study design, data collection, and interpretation, or the decision to submit the work for publication.

We thank many laboratories for providing the plasmids.

© 2022, Park et al.

This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.