The first pathogenic mutation in CaV1.2 was identified in 2004 and was shown to cause a severe multisystem disorder known as Timothy syndrome (TS). The mutation was localized to the distal S6 region of the channel, a region known to play a major role in channel activation. TS patients suffer from life-threatening cardiac symptoms as well as significant neurodevelopmental deficits, including autism spectrum disorder (ASD). Since this discovery, the number and variety of mutations identified in CaV1.2 have grown tremendously, and the distal S6 regions remain a frequent locus for many of these mutations. While the majority of patients harboring these mutations exhibit cardiac symptoms that can be well explained by known pathogenic mechanisms, the same cannot be said for the ASD or neurodevelopmental phenotypes seen in some patients, indicating a gap in our understanding of the pathogenesis of CaV1.2 channelopathies. Here, we use whole-cell patch clamp, quantitative Ca2+ imaging, and single channel recordings to expand the known mechanisms underlying the pathogenesis of CaV1.2 channelopathies. Specifically, we find that mutations within the S6 region can exert independent and separable effects on activation, voltage-dependent inactivation (VDI), and Ca2+-dependent inactivation (CDI). Moreover, the mechanisms underlying the CDI effects of these mutations are varied and include altered channel opening and possible disruption of CDI transduction. Overall, these results provide a structure–function framework to conceptualize the role of S6 mutations in pathophysiology and offer insight into the biophysical defects associated with distinct clinical manifestations.
CaV1.2 L-type Ca2+ channels are perhaps the most prevalent of the voltage-gated Ca2+ channels, existing in cardiac, neuronal, and smooth muscle cells (Adams and Snutch, 2007). Ca2+ entry through these channels is precisely controlled through multiple forms of regulation including voltage-dependent activation, voltage-dependent inactivation (VDI), and Ca2+-dependent inactivation (CDI; Lee et al., 1985; Peterson et al., 1999; Zuhlke et al., 1999; Liang et al., 2003; Beyl et al., 2009). Each of these regulatory processes is vital, and their disruption is expected to produce adverse physiological consequences. Timothy syndrome (TS) represents one such class of mutations, in which a single point mutation within CaV1.2 leads to a severe multisystem disorder characterized by developmental delays, autism spectrum disorder (ASD), and profound long-QT syndrome (LQTS; Napolitano et al., 1993; Splawski et al., 2004; Splawski et al., 2005). The underlying cause of TS was first described in 2004 as a single point mutation (G406R) within the distal IS6 region of CaV1.2 (Splawski et al., 2004). This form of LQTS is among the most severe presentation of the syndrome, with patients exhibiting QT intervals between 480 and 730 ms (Marks et al., 1995a; Marks et al., 1995b; Splawski et al., 2004; Splawski et al., 2005). As a result, patients suffer from life-threatening arrhythmias, which are often fatal in early childhood. Furthermore, TS represents one of the most penetrant monogenic forms of ASD (Splawski et al., 2004; Splawski et al., 2005; Lu et al., 2012). Since the first discovery of the G406R mutation, a growing number of mutations have been identified in CaV1.2, most of which result in significant cardiac phenotypes (Splawski et al., 2004; Splawski et al., 2005; Gillis et al., 2012; Wemhoner et al., 2015; Landstrom et al., 2016). However, only select mutations also cause neurological phenotypes, often including ASD or developmental delay (Splawski et al., 2004; Splawski et al., 2005; Gillis et al., 2012; Boczek et al., 2015a; Wemhoner et al., 2015). As the number and variety of identified CaV1.2 mutations continue to expand, so too does the phenotypic variation observed in patients. In the heart, the importance of CaV1.2 inactivation has long been recognized and there is a clear mechanistic link between excess Ca2+ entry through CaV1.2 and prolongation of the cardiac action potential (AP; Alseikhan et al., 2002; George, 2015; Morales et al., 2019), the cellular hallmark for LQTS. However, the mechanistic link between CaV1.2 mutations and neurological effects remains much less clear. Mutations within both CaV1.2 and CaV1.3 channels which cause large effects on channel gating have been linked to ASD; however, some CaV1.2 mutations with significant effects on channel gating produce no neurological phenotype, and simple gain-of-function/loss-of-function descriptions of CaV1.2 mutations result in little correlation with patient phenotypes (Liao and Soong, 2010; Wemhoner et al., 2015; Ozawa et al., 2018).
The original G406R TS mutation is located within the distal IS6 region of the channel (Splawski et al., 2004), as is the G402S mutation, which was identified soon after (Splawski et al., 2005). Both mutations produce a marked deficit in VDI (Splawski et al., 2005; Dick et al., 2016), fitting with prior work demonstrating that residues within the distal S6 region may be critical to the function of a “hinged-lid” inactivation scheme (Stotz et al., 2000; Stotz and Zamponi, 2001b; Tadross et al., 2010). Moreover, the S6 is known to play a critical role in channel activation (Raybaud et al., 2006; Kudrnac et al., 2009; Tadross et al., 2010; Hering et al., 2018). It is therefore not surprising that each of these mutations alters the voltage dependence of channel activation, although in opposing directions (G406R, hyperpolarizing shift; G402S, depolarizing shift; Dick et al., 2016). Finally, both mutations have been shown to cause a decrease in CDI, demonstrating multiple effects on channel gating. Interestingly, the effect on CDI for many mutations within this region of both CaV1.2 (including G402S and G406R) and CaV1.3 has been demonstrated to be a direct result of a change in channel activation (Tadross et al., 2010; Dick et al., 2016), implying that the S6 region of the channel may not play a direct role in CDI. While a growing number of CaV1.2 mutations have been identified across multiple channel regions, a large subset of these pathogenic mutations appears within distal S6 regions of the channel. Thus, there may be common mechanistic elements that can be elucidated by evaluating the effects of these S6 mutations.
Here, we evaluate the biophysical impact of select CaV1.2 mutations on CDI, VDI, and channel activation, focusing primarily on mutations near the distal S6 region of the channel. Despite the locus of these mutations near the channel activation gate, we find that not all S6 mutations exert an effect on channel activation. In contrast to the canonical TS mutations, we find that the S6 mutations within this study have more selective effects on channel properties, sometimes altering CDI or VDI independently. Further investigation into the mechanisms underlying the gating changes in CDI-altering mutations reveals a remarkably diverse impact of S6 mutations. In particular, two mutations (I1166V and I1166T) caused a marked change in single-channel properties indicative of enhanced entry into mode 2 gating, while another distal S6 mutation (I1475M) resulted in a selective CDI deficit, implicating a greater role for the IVS6 region in CDI. Finally, we show that these biophysical changes produce a difference in the response of the channels to a neurological stimulus, demonstrating that the neurological phenotype may be most strongly associated with enhanced channel activation.
Materials and methods
The α1C backbone used in this study was the previously described human cardiac CaV1.2 clone (Dick et al., 2016) within the pcDNA3 plasmid and corresponds to GenBank accession no. Z34810. The original plasmid used to generate this clone was a kind gift from Tuck Wah Soong (Tang et al., 2004). Channel mutations were introduced into this plasmid using the QuickChange Lightning kit from Agilent. All portions of the resulting construct that were subject to PCR were confirmed by DNA sequencing.
A homology model of CaV1.2 was generated based on the 5GJW PDB structure of the CaV1.1 channel (Wu et al., 2016). Sequences of the human CaV1.2 channel used in this study and the CaV1.1 channel corresponding to the PDB structure were aligned. The CaV1.2 sequence was trimmed to match the resolved regions of the CaV1.1 structure. Fortunately, all mutations described in this study fell within a resolved region of the channel. Following alignment and trimming, the software Modeller was used to generate the model (Webb and Sali, 2016). The subsequent structural model was visualized using PyMol, and figures were generated from this software.
All patch clamp experiments were performed using HEK 293 cells transfected via the calcium phosphate method (Brody et al., 1997). Cells were transfected with the human α1C subunit along with the rat α2δ subunit (GenBank accession no. NM_012919.2) and either rat brain β1b or β2a (Perez-Reyes et al., 1992) as indicated. The SV40 T antigen was also cotransfected to enhance expression levels. To facilitate the identification of transfected cells, all transfections included GFP either as a separate plasmid (for β1b) or as part of a GFPIR vector containing the beta subunit (β2a). Cells were transfected and used for patch clamp recordings within 1–3 d.
Whole-cell patch clamp
Whole-cell patch clamp recordings were performed at room temperature using an Axopatch 200B amplifier (Axon Instruments). Borosilicate glass electrodes were generated with resistances between 1 and 3 MΩ, with series resistance compensated to >70% during the recordings. Currents were low-pass filtered at 2 kHz (4-pole Bessel filter) and sampled at 10 kHz during VDI and CDI measurements. Tail current protocols and neuronal AP recordings were recorded with 5 kHz low-pass filtering and sampled at 50 kHz so as to enable full resolution of the peak. For activation curve, CDI, VDI, and Bay K 8644 response measurements, the internal solution contained (in mM) 114 CsMeSO3, 5 CsCl, 4 MgATP, 10 HEPES (pH 7.4), and 10 BAPTA (1,2-bis[o-aminophenoxy]ethane-N,N,N′,N′-tetraacetic acid) at 295 mOsm adjusted with CsMeSO3, and the bath solution contained (in mM) 102 TEA-MeSO3, 10 HEPES (pH 7.4), and 40 CaCl2 or BaCl2 at 305 mOsm adjusted with TEA-MeSO3. Solutions for the neuronal stimulus were adjusted to more closely approximate physiological Ca2+ such that the internal solution contained (in mM) 114 CsMeSO3, 5 CsCl, 4 MgATP, 10 HEPES (pH 7.4), and 1 EGTA (ethylene glycol-bis[β-aminoethyl ether]-N,N,N′,N′-tetraacetic acid) at 295 mOsm adjusted with CsMeSO3, and the bath solution contained (in mM) 102 TEA-MeSO3, 40 HEPES (pH 7.4), and 1.8 CaCl2 at 305 mOsm adjusted with TEA-MeSO3. Data were analyzed using custom MATLAB scripts with all fits and statistical analysis done in Prism (GraphPad). Inactivation was quantified as the ratio of current remaining after 300 ms in either Ca2+ or Ba2+ (r300), such that the difference between the r300 values recorded in Ba2+ versus Ca2+ quantified pure CDI. For Bay K 8644 experiments, 50 mM stock solutions of (±) Bay K 8644 (catalog #: B112; Sigma-Aldrich) were made in DMSO and freshly diluted to a final 5 µM concentration in external solution on the day of recording. Quantification of the current increase due to Bay K 8644 was determined as the relative increase in current amplitude during a 300-ms step to 20 mV (i20).
Single channel recordings
Single channel patch clamp recordings were performed at room temperature using an Axopatch 200B amplifier (Axon Instruments). Thick-walled borosilicate glass electrodes were generated with resistances between 3 and 5 MΩ and coated with Sylgard. Currents were low-pass filtered at 2 kHz (4-pole Bessel filter) and sampled at 10 kHz. The pipette solution matched the whole-cell extracellular solution and contained (in mM) 102 TEA-MeSO3, 10 HEPES (pH 7.4), and 40 BaCl2 at 305 mOsm adjusted with TEA-MeSO3. The bath solution was designed to zero the membrane potential and contained (in mM) 132 K-glutamate, 5 KCl, 5 NaCl, 3 MgCl, 2 EGTA, 10 glucose, and 20 HEPES (pH 7.4) at 300 mOsm adjusted with glucose.
The ramp protocol was applied from −80 to 70 mV over a duration of 200 ms. The leak for each sweep was manually fit using a combination of linear and exponential functions. The unitary current amplitude was fit with the GHK equation (Hille, 2001), providing a readout for conductance. The voltage parameter Vs was allowed to vary by ±3 mV between patches to allow for recording variability. For each cell, traces were averaged, excluding blank traces, to produce a current–voltage relation. These curves were then averaged together for different patches and the PO was determined using the GHK relation. Prior to the conclusion of the experiment, Bay K 8644 was added to a final concentration of ∼5 µM to allow for easy counting of the number of channels in the patch. We found that our estimate of the number of channels was consistent with the count in Bay K 8644, therefore recordings in which the cell died prior to Bay K 8644 addition (but with >50 sweeps recorded) did not need to be excluded from the average. Patches with 1–3 channels were included in the average, and blanks were excluded such that the data corresponds to active sweeps only, as previously described for CaV1.2 channels harboring TS mutations (Dick et al., 2016).
For simultaneous Ca2+ uncaging and patch clamp recordings, we used a calibrated mix of 5 μM Fluo-2 (TefLabs), 5 μM Fluo-2 Low Affinity (TefLabs), and 2.5 μM Alexa568 (Invitrogen) as previously described (Lee et al., 2015; Limpitikul et al., 2018). This combination enabled ratiometric recording across a range of Ca2+ concentrations. Stocks of this solution were premixed and calibrated prior to the experiments so that there was no variability in the ratios across experiments. In addition to these dyes, the internal solution contained (in mM) 135 CsCl, 40 HEPES (pH 7.4), 1–4 DMNP-EDTA, 1–20 Citrate, and 0.75–3.5 CaCl. Citrate, DMNP, and CaCl were adjusted to provide varying levels of Ca2+ steps, such that baseline calcium levels were below 100 nM as measured by the Ca2+ dyes. External solutions contained (in mM) 70 TEA-MeSO3, 10 HEPES (pH 7.4), and 40 CaCl. Data were analyzed by custom MATLAB software (MathWorks), and final fits to the data were done in Prism (GraphPad).
Online supplemental material
Pathogenic CaV1.2 mutations often appear near the distal S6
The first TS mutations described (G406R and G402S) were both located within the IS6 region and caused changes to channel activation (Splawski et al., 2004; Splawski et al., 2005; Dick et al., 2016). Since this first description, numerous additional mutations in the proximity of the distal S6 of CaV1.2 have been reported (Boczek et al., 2015a; Wemhoner et al., 2015; Landstrom et al., 2016; Marcantoni et al., 2020). Given the importance of this region in channel regulation, we chose several pathogenic mutations near a distal S6 region of the channel to focus on. Fig. 1 A displays the locus of each mutation on a cartoon of the CaV1.2 channel, with the S6 regions highlighted in blue for easy reference. Mutation L762F (Landstrom et al., 2016) resides just past the IIS6 region and has been shown to cause a cardiac selective phenotype (Table S1). We next considered two mutations within the IIIS6 region. Interestingly, I1166T and I1166V (Boczek et al., 2015a; Wemhoner et al., 2015) occur in the identical residue, yet produce distinct phenotypes, with I1166T producing a multisystem disorder often including neurological effects, while I1166V causes a cardiac selective effect (Table S1). From the IVS6 domain, we focused on the mutation I1475M, which has been described as cardiac selective in patients (Wemhoner et al., 2015). Finally, we considered the effect of one mutation which resides outside an S6 region. E1496K resides within the EF-hand located on the C-tail of the channel and is associated with a cardiac-selective phenotype (Wemhoner et al., 2015). This mutation is included in this analysis as this region has previously been identified as critical for VDI (Kim et al., 2004; Raybaud et al., 2006), and has been suggested to cause a shift in channel activation (Wemhoner et al., 2015) similar to the effects seen with S6 mutations.
To fully appreciate the location of the mutations on the channel, we generated a homology model of CaV1.2 based on the structure of CaV1.1 (Wu et al., 2016; Zhao et al., 2019), enabling visualization of the mutation loci and the surrounding residues (Fig. 1 B). The S6 regions are highlighted in blue, and the visualization of the model structure with only the S6 regions displayed clearly demonstrates the location of the mutations near the helical bundle that forms the channel activation gate, with the exception of the E1496K mutation which resides within the EF-hand, highlighted in green. Interestingly, while mutation L762F is predicted to lie just distal to the IIIS6 based on sequence, the homology model predicts that it does indeed reside on the S6 helix, which is extended in this domain.
Given the location of these mutations, we expect many may have effects on channel activation, as we previously showed for G406R and G402S (Dick et al., 2016). We therefore utilized a tail-current protocol (Dick et al., 2016) to define the relative activation curve for each mutation (Fig. 1, C–H). Interestingly, despite its locus on the IIS6 helix, L762F did not significantly change the V1/2 of activation (Fig. 1 D). However, when we explored the effect of I1166V, we found a small left shift in channel activation which was not of statistical significance (P = 0.73). Evaluation of I1166T revealed a much larger left shift, demonstrating the critical nature of this locus in channel activation. Next, we measured the effects of I1475M and E1496K and found that neither caused a change in the V1/2 of activation.
In addition to altered channel activation, prior studies have indicated that the canonical G406R mutation also slows the kinetics of channel deactivation (Yarotskyy et al., 2009). We therefore evaluated the rate of deactivation of each mutant. To do this, we utilized the tail current data obtained from an 80 mV pre-pulse so as to permit maximal opening of channels (Fig. 1 I). Indeed, both I1166V and I1166T produced a marked slowing of channel deactivation, which was quantified by a double exponential fit. Both the fast and slow τ components were significantly slowed by each mutation at the I1166 residue, with the T substitution producing the largest effect.
CaV1.2 mutations have selective effects on channel inactivation
Both canonical TS mutations have significant effects on channel inactivation, nearly eliminating VDI and blunting CDI (Dick et al., 2016). These effects have each been shown to have a clear impact on the cardiac AP consistent with the LQTS phenotype of the patients (Morotti et al., 2012; Dick et al., 2016; Morales et al., 2019). We therefore considered the impact of each of the mutations in this study on VDI and CDI. We begin by looking at VDI, which can be quantified through the use of a β subunit permissive of VDI, such as β1b (Wei et al., 2000), with Ba2+ as the charge carrier so as to remove the confounding effect of Ca2+-dependent processes such as CDI. Under these conditions, VDI can be viewed as the decay in current during a 300-ms step depolarization (Fig. 2 A, blue shade). Quantification of VDI is given by the metric r300, which is the ratio of current remaining after 300 ms as compared to peak current such that a value of 1 would equate to no VDI, while 0 would indicate complete inactivation. Plotting r300 for WT CaV1.2 demonstrates the expected increase in VDI as a function of voltage (Fig. 2 A, right). We next considered whether a VDI deficit may underlie the phenotype of patients harboring one of the mutations which lacked an activation shift. Indeed, the evaluation of L762F demonstrated a near complete loss of VDI (Fig. 2 B), similar to the VDI deficit identified for both G406R and G402S. Looking at E1496K (Fig. 2 C), we saw a significant decrease in VDI, although to a lesser extent as compared to L762F. However, when we evaluated I1475M, we found that VDI was preserved (Fig. 2 D), leaving the mechanism underlying the patient phenotype for this mutation unresolved. Finally, we evaluated the mutations I1166V and I1166T, which altered channel activation; however, VDI remained unperturbed compared with WT (Fig. 2, E and F).
We next considered the effect of each mutation on CDI. To do this, we utilized the β2a subunit, which has been shown to minimize VDI (Wei et al., 2000), thus enabling the evaluation of CDI in relative isolation of the confounding effects of altered VDI. A minimal amount of residual VDI can be seen in the Ba2+ current trace (Fig. 2 G, black). However, when Ca2+ is used as the charge carrier through the channel (Fig. 2 G, red), robust inactivation can be seen, and CDI can be defined as the difference between the inactivation in Ca2+ as compared to Ba2+ (peach shaded area). Again, the metric r300 quantifies the extent of inactivation in Ca2+ (red) and Ba2+ (black), while f300 is quantified as the difference between the Ca2+ and Ba2+r300 values and provides a metric of pure CDI. Repeating this measurement for the L762F and E1496K mutations demonstrated no effect on CDI (Fig. 2, H and I), consistent with the lack of change in channel activation. However, evaluation of I1475M, which also displayed normal activation parameters, revealed a significant decrease in CDI across multiple voltages (Fig. 2 J). Thus, this mutation deviates from previous L-type channel mutations which caused a change in CDI secondary to an activation change (Tadross et al., 2010; Dick et al., 2016). Finally, I1166V and I1166T (Fig. 2, K and L) also displayed a significant decrease in CDI, with I1166T again displaying a more severe effect. Overall, evaluation of these mutations demonstrates somewhat selective effects on channel regulation, with separable effects on VDI, CDI, and activation.
Reduced CDI at maximal Ca2+ concentrations revealed by Ca2+ uncaging
Our previous work has demonstrated that a reduction in CDI can occur through more than one mechanism (Dick et al., 2016). We therefore wanted to look more closely at the underlying deficits in the three CDI deficient channels. To do this, we considered an allosteric model for CaV1.2 gating in which channels initially open in a Ca2+-free configuration with a relatively high open probability (PO), known as mode 1 (Hess et al., 1984). Upon Ca2+ binding to calmodulin (CaM), channels transition into a low PO state (mode Ca; Imredy and Yue, 1994), resulting in CDI. Within this scheme, CDI, which is viewed at the whole-cell level consists of two components: FCDI is the fraction of channels that enters mode Ca and is dependent on the Ca2+ driving force, while CDImax is the maximum amount of CDI you would get if all channels were to enter mode Ca2+ and is dependent on the difference between the PO of mode 1 versus mode Ca (Fig. 3 A; Tadross et al., 2010). Our prior work has demonstrated that pathogenic CaV1.2 mutations can exert an effect on CDI by changing either of these two components (Dick et al., 2016). We therefore sought to measure the effect of CDI-altering mutations on CDImax. To do this, we utilized Ca2+ photouncaging paired with simultaneous patch clamp recordings and Ca2+ imaging. Within this setup, our goal was to drive CDI with a known concentration of Ca2+, enabling us to determine the extent of CDI as a function of Ca2+ concentration. DM-Nitrophen (DMNP) was used as the Ca2+ cage and added to our intracellular solution along with the Ca2+ indicator dye Flo4-FF. We could then release Ca2+ into the cytosol of the cell using a UV flash (Fig. 3 A), while quantifying the amount of Ca2+ released through fluorescent imaging (Dick et al., 2016).
Our setup is similar to our prior study in which we characterized CaV1.2 CDI as a function of Ca2+ (Dick et al., 2016); however, in that case, we utilized a mutant CaV1.2 channel to prevent intracellular Ca2+ release from causing a pore-block (Cloues et al., 2000). However, the ability to quantify Ca2+-dependent regulation in WT Ca2+ channels has since been demonstrated for CaV2.1 channels (Lee et al., 2015). Given the potential for a channel mutation to impact our results, we opted to adapt this method for unaltered CaV1.2 channels. To do this, we utilized Ca2+ as the charge carrier, thus bypassing the pore block, which would be induced following Ca2+ uncaging in the presence of an alternative charge carrier (Dick et al., 2016). Next, we ensured that the cytosolic Ca2+ concentration would represent the Ca2+ concentration at the mouth of the channel by minimizing Ca2+ entry through the channel. Specifically, we applied a short step to the reversal potential of the channel (90 mV), which allowed channels to open without significant entry of Ca2+. Tail currents were then measured during a subsequent step to −40 mV, which was chosen so as to allow channels to fully close with kinetics enabling robust resolution of peak tail currents. A UV flash was then applied, elevating intracellular Ca2+, and the tail current protocol was repeated. Under these conditions, CDI was quantified as the difference in peak tail current following uncaging relative to current prior to Ca2+ release (Fig. 3 B, left). Quantitative Ca2+ imaging then provided a readout of the cytosolic Ca2+ concentration for each tail current measured (Fig. 3 B, red), and repeating the protocol for variable amounts of Ca2+ released enabled the generation of a relation between CDI and cytosolic Ca2+ (Fig. 3 B, right). Importantly, repeated application of the tail current protocol produced stable current amplitudes and did not alter the measured cytosolic Ca2+ concentration. For WT channels, we obtained a CDI curve with a KD = 1.1 ± 0.04 µM, Hill coefficient = 2, and CDImax = 0.75 ± 0.02, a result nearly identical to our prior measurements with the mutant channel (Dick et al., 2016).
We next applied our Ca2+ uncaging protocol to the channels harboring CDI-altering mutations. Our previous work has demonstrated that a left shift in channel activation can be causative of a decrease in CDImax (Dick et al., 2016). We therefore began by considering whether the modest left shift in activation due to the I1166V mutation also caused a reduction in CDImax. Indeed, application of our tail current protocol to channels harboring the I1166V mutation (Fig. 3 C) demonstrated a significant decrease in CDImax (KD = 1.4 ± 0.15, Hill coefficient = 2, and CDImax = 0.56 ± 0.05). We next considered the effect of the I1166T mutation, which produced a large left shift in channel activation. Likewise, this mutation resulted in a robust decrease in CDImax (KD = 1.27 ± 0.08, Hill coefficient = 2, and CDImax = 0.48 ± 0.02). Finally, we considered the effect of the I1475M mutation, which had no shift in channel activation. Despite this lack of activation change, we found that this mutation also caused a decrease in CDImax (KD = 1.1 ± 0.09, Hill coefficient = 2, and CDImax = 0.57 ± 0.02). Given the lack of shift in the activation curve for this mutation, this result indicates that the loss of CDI proceeds from a distinct mechanism not previously described for CaV1.2 mutations.
Single-channel properties reveal a mode-switch in CaV1.2 mutant channels
All CDI-altering mutations in this study demonstrated a decrease in CDImax (Fig. 3), a metric dependent on the PO of the channel (Tadross and Yue, 2010). FCDI, on the other hand, cannot be directly measured but is determined by the Ca2+ entering the channel during each opening. It is therefore dependent on single-channel parameters including channel conductance and PO. Therefore, to fully understand the mechanism underlying the CDI deficit, we undertook single-channel recordings for each CDI-altering mutation. For these experiments, Ba2+ was used as the charge carrier, and channels were expressed with the β2a subunit to facilitate the evaluation of single channel properties in the absence of inactivation (Wei et al., 2000). Application of a voltage ramp across a single CaV1.2 channel resulted in channel openings with a conductance of 0.018 pA mV−1 (Fig. 4 A), as previously recorded for this channel under these experimental conditions (Dick et al., 2016). Averaging numerous sweeps across multiple cells enabled the resolution of a PO curve (Fig. 4 B, black), which could be well fit by a Boltzmann distribution (red). Moreover, we measured the individual open durations during the ramp and found that channels consistently exhibited brief openings, previously described as mode 1 gating (Fig. 4 C; Hess et al., 1984). After collecting hundreds of sweeps, we added Bay K 8644, which is known to induce mode 2 gating (Fig. 4 A, gray box), in which channels open at more negative potentials and with significantly longer open durations (Hess et al., 1984). This maneuver ensures accuracy when counting the number of channels within each patch.
We next applied the ramp protocol to channels harboring the I1166V mutation, which produced a modest left shift in channel activation in whole-cell experiments (Fig. 1 E). Indeed, single-channel measurements recapitulated the left shift in activation (Fig. 4, D and E), without significant overall changes to conductance or the maximal PO (PO,max). Interestingly, the left shift in activation was primarily due to periodic entry into mode 2. While WT CaV1.2 channels will occasionally enter mode 2 without perturbation, the general application of an agonist such as Bay K 8644 (Fig. 4, gray box) is required to cause mode 2 gating. The mode 2 gating caused by I1166V is demonstrated in the exemplar traces displayed (Fig. 4 D), where sweeps from the same recording switch from a mode 1 characteristic (top two traces) to mode 2 (bottom trace). This is further quantified by the open duration histogram (Fig. 4 F), which displays an enhanced number of long-duration openings. This mode 2 switching was even more significant when we looked at the I1166T mutation (Fig. 4 G). There was again no change in channel conductance; however, channels harboring the I1166T mutation spent a significant amount of time in mode 2, resulting in a large left shift in channel activation as well as an overall increase in PO,max (Fig. 4 H). Open duration histograms demonstrate a considerable shift to long duration openings, with durations extending beyond 40 ms (Fig. 4 I). This remarkable behavior results in a substantial increase in channel opening across multiple voltages and correlates with the severity of the patient phenotype. Interestingly, while this increased PO explains the reduction in CDImax (Fig. 3 D), it also points to an increase in FCDI at lower voltages as additional Ca2+ would now be available to drive channels into mode Ca. Thus, the effects of each mutation on these two parameters are unlikely to be entirely independent. Moreover, increased entry into mode 2 is consistent with a decrease in channel deactivation (Fig. 1 I) as Bay K 8644 has been shown to produce the same pattern of gating (Yarotskyy et al., 2009).
Finally, we turned our attention to the I1475M mutation, which displayed a loss of CDI through a deficit in CDImax despite no change in channel activation. Surprisingly, this mutation produced minimal changes to the single channel properties, with no significant change in conductance, V1/2 of activation, or open duration (Fig. 4, J–L). While we did note a slight decrease in PO,max (Fig. 4 K), this small effect cannot account for the CDI changes produced by this mutation. It therefore appears that I1475M produces a loss of CDI through a novel mechanism.
Diminished sensitivity of mutant channels for Bay K 8644
As the I1166T and I1166V mutant channels exhibit increased propensity for mode 2 behavior, they may have reduced sensitivity to upregulation by the CaV1.2 agonist Bay K 8644, which acts on the channels largely through an increase in mode 2 gating (Hess et al., 1984). WT CaV1.2 channels displayed the expected large increase in Ba2+ current amplitude in response to 5 µM Bay K 8644 (Fig. 5 A). Quantifying the average data as a function of voltage also demonstrated the expected left shift in voltage dependence (Hess et al., 1984). To compare the increase in current across channels with differing basal current densities, we normalized all data to the peak current prior to the application of Bay K 8644 and defined the metric i20 as the increase in amplitude due to the application of Bay K 8644 during a 20-mV step depolarization. Applying the same protocol to channels harboring the I1166T mutation demonstrated a significantly smaller increase in current amplitude (Fig. 5 B). Interestingly, the left shift in peak current in I1166T exceeded that of the WT channels in the presence of Bay K 8644. Application of Bay K 8644 induced an even further left shift on the mutant channel, indicating that the altered voltage dependence of the I1166T-containing channels may precede a mechanism independent of mode 2 initiation. Application of Bay K 8644 to the I1166V mutation resulted in a similar trend toward reduced response (Fig. 5 C), fitting with the modest enhancement of basal mode 2 gating; however, the i20 value was not statistically different from WT (P = 0.13). Likewise, I1475M displayed no statistical difference between mutant and WT metrics (Fig. 5 D).
A closer look at the neuronal effects of CaV1.2 mutations
The majority of CaV1.2 mutations reported in patients, and all in this study, are associated with cardiac symptoms (Boczek et al., 2015b; Marcantoni et al., 2020). However, only a select few are associated with the severe neurological phenotypes identified among TS patients (Boczek et al., 2015b; Marcantoni et al., 2020). While alterations in CDI, VDI, and activation represent a recognized mechanism for inducing LQTS (Morotti et al., 2012; Morales et al., 2019), the same cannot be said for the developmental delay or ASD seen in some CaV1.2 channelopathy patients. Of the mutations evaluated in this study, only I1166T has been described as causing a neurological deficit in patients. Interestingly, this mutation also results in a dramatic increase in channel activation, both through a left shift in the voltage-dependence of activation and enhanced entry into mode 2 gating (Fig. 4 H). We therefore wanted to probe whether these biophysical properties were sufficient to alter the response of the channels to a neuronal action potential. By expressing the channels within HEK 293 cells and applying a neuronal electrical stimulus, we observe only those effects due directly to channel gating, eliminating the contribution of downstream signaling, channel trafficking, or developmental changes. To more closely match physiological conditions, experiments were carried out in an external solution containing 1.8 mM Ca2+ and an internal solution containing 1 mM EGTA, which permits more physiological intracellular Ca2+ buffering (Stern, 1992). In addition, we utilized the β1b subunit such that both CDI and VDI are enabled (Wei et al., 2000). Ca2+ currents were then recorded in response to a 40-Hz train of neuronal APs and the amplitude of each Ca2+ response was measured (Fig. 6 A). For WT channels, the peak current in response to the repetitive APs decayed slightly over the course of the stimulus. However, when CaV1.2 channels harboring the I1166T mutation were exposed to the same stimulus, the amplitude decay was significantly increased (Fig. 6 B). To quantify this result across cells, we normalized the current amplitude by the first peak (Fig. 6 C) and found that the fraction of current remaining after 20 APs was significantly less for the I1166T mutation as compared with WT channels (Fig. 6 D). The same protocol applied to I1166V, I1475M, L762F, or E1496K resulted in amplitude changes nearly identical with the WT channels (Fig. 6, E–H). Thus, unlike mutations that selectively alter CDI or VDI, the gating changes of the I1166T mutation are sufficient to produce a differential response to a neuronal signal. As G406R has also been shown to cause a left shift in channel activation (Dick et al., 2016) in addition to significant deficits in CDI and VDI, we considered how this mutation would respond to the neuronal AP stimulus train. Indeed, similar to I1166T, channels harboring the G406R mutation exhibited and enhanced the decay of peak current amplitude during the stimulus train (Fig. 6 I). Interestingly, the magnitude of the effect was smaller than that of the I1166T, fitting with the smaller shift in channel activation of the G406R mutation (Table S1). Moreover, the response to the neuronal AP train was highly frequency dependent, with both G404R and I1166T deviating from the WT response only at frequencies of 20 Hz or higher (Fig. S1).
Finally, we considered whether the altered deactivation kinetics of select mutations might impact the morphology of the current response to the AP stimulus. To this end, we measured the half-width of the current for each mutant channel and found that indeed, the I1166T and G406R mutations extended the duration of the current response (Fig. 6 J).
While these results indicate a unique feature for the I1166T and G406R mutations, they also appear to violate the prior results which showed a loss of inactivation for both mutations. Here, both I1166T and G406R appear to enhance the inactivation of the current in response to a train of APs. However, under the conditions of this experiment, it is unlikely that maximal channel inactivation is achieved during the short duration of the neuronal APs, even after repetitive stimulation. Instead, we postulated that the mutations cause a large increase in the amplitude of the initial peak; however, the need to normalize the data to compare across cells removed this component. The subsequent enhanced decay of the current through channels harboring the mutations would then result from an augmented Ca2+ driving force (increased FCDI). Thus, the effects of I1166T are due to a complex interplay between a deficit in CDImax (Fig. 3 D) and an increase in FCDI. To demonstrate this, we considered that the effect of the I1166T mutation is very similar to the effect of Bay K 8644 when applied to WT channels (Fig. 4, A and G; Hess et al., 1984). This allowed us to mimic the impact of the gating changes associated with this mutation within the same cell. Indeed, when the AP train was applied to WT cells before and after the application of Bay K 8644, a clear increase in current amplitude due to Bay K 8644 application can be seen (Fig. 6 K). The enhanced decay of the current then brings the amplitude back down toward the baseline. Thus, the cumulative effect of the I1166T mutation on CaV1.2 current amplitude is likely that of a gain-of-function, as predicted by the increased activation of the mutant channel. Finally, we evaluated the halfwidth of the Bay K 8644–treated cells and found that like the I1166T, Bay K 8644 increased the duration of the current response (Fig. 6 L), fitting with the known slowing of deactivation kinetics associated with Bay K 8644–induced stabilization of mode 2 (Tsien et al., 1986).
Mutations in CaV1.2 have been linked to severe cardiac and neurological disorders, which are often fatal in early childhood. The first two mutations described, G406R and G402S, produced profound phenotypes in patients despite the fact that they occurred within a mutually exclusive exon, resulting in expression in only a fraction of CaV1.2 channel variants (Splawski et al., 2004; Splawski et al., 2005). This severe phenotype at low expression levels may be partly due to the fact that the mutations cause significant changes to channel activation, VDI, and CDI (Barrett and Tsien, 2008; Dick et al., 2016). The mutations evaluated in the current study, however, are each expressed in a constitutive exon, resulting in much higher expression in the patient. Thus, the more selective effects on channel regulation are consistent with the patient phenotypes; severe mutations such as G406R would likely be embryonic lethal at these higher expression levels (Dick et al., 2016). Nonetheless, the gating effect of the I1166T mutation is quite severe, even in the absence of a VDI deficit. This observation fits with the relative severity of this mutation in patients, where I1166T produced the most severe phenotype of those in this study (Boczek et al., 2015a; Wemhoner et al., 2015; Landstrom et al., 2016). It is also the only mutation in this study that correlates with a severe multisystem disorder similar to that described in the original TS patients harboring the G406R mutation (Table S1). While I1166T did not produce the 100% penetrance of ASD described in the original cohort of TS G406R patients, of the two studies describing this mutation, neurodevelopmental delays were identified in one (Wemhoner et al., 2015), progressive cerebral and cerebellar atrophies and intellectual impairment were identified in the second (Boczek et al., 2015a; Wemhoner et al., 2015), and seizures were identified in both. Thus, neurological involvement may be a component of a more complex etiology for this mutation.
Thus far, attempts to correlate biophysical changes due to CaV1.2 mutations and patient phenotype have not yielded definitive results (Landstrom et al., 2016). This may be partly due to the comparison of experimental results across multiple conditions from different labs or from the limited number of in-depth mechanistic studies that distinguish between different forms of channel regulation. It has also been suggested that neuronal effects of some CaV1.2 mutations may be independent of Ca2+ permeation (Krey et al., 2013). We therefore sought to elucidate the detailed changes in channel gating due to CaV1.2 mutations. We evaluated several mutations at both the whole-cell level (with conditions tuned for robust measurements of VDI versus CDI) and further examined the mechanism underlying the CDI-altering mutations through quantitative Ca2+ imaging and single channel recordings. Within these parameters, we do indeed find that only the multisystem I1166T mutation produces a large increase in channel activation. Moreover, our results indicate that the gating changes of this mutation are sufficient to reproduce a change in neurological response even in the absence of downstream signaling or neuron-specific regulation (Fig. 6). Interestingly, the differential response of the mutant channels only occurs at higher frequency AP trains, consistent with previous studies which implicate alterations in neuronal burst firing in neurodevelopmental disorders (Calorio et al., 2019; Lee et al., 2021). Moreover, both G406R and I1166T share these features of enhanced channel activation and neurological phenotypes and behave similarly in our neuronal AP assay. On the other hand, CDI and VDI deficits are present in multiple cardiac-selective mutations which did not produce an overt change in the current response to a neuronal stimulus. We therefore postulate that mutations that decrease CDI or VDI or enhance channel activation are causative of LQTS; however, only those with significant enhancement of channel activation appear likely to result in ASD or neurodevelopmental disorders (Fig. 7). The importance of enhanced channel activation due to the G406R mutation has previously been described as a likely contributor to increased basal transcriptional activation (Servili et al., 2020) and is postulated to represent a critical component of the neurological pathology of TS (Calorio et al., 2019; Marcantoni et al., 2020). Interestingly, this mechanism would be consistent with the mistuning of a voltage-dependent conformational change, which has been previously identified as important for the pathogenesis of canonical TS mutations (Li et al., 2016). Thus, in a neuron, the full impact of the TS mutations is likely to proceed from more than one pathogenic mechanism, with enhanced activation as a common feature (Marcantoni et al., 2020).
In addition to providing insight into the mechanisms underlying the pathogenesis of CaV1.2 mutations, this study also expands our knowledge of the role of the S6 region in channel gating. While pathogenic mutations in CaV1.2 have been identified across multiple distinct channel regions, many appear to cluster near a distal S6 and exert their impact through a change in channel gating. Interestingly, this S6 pattern is emerging as a common theme across multiple ion channels (Splawski et al., 2005; Shelley et al., 2013; Biel et al., 2016; Kschonsak et al., 2020; Lory et al., 2020), making S6-opathies an important and growing class of ion channel diseases (Lory et al., 2020). All but one of the mutations in this study occurs within a distal S6 region, and our results demonstrate that these S6 mutations can exert independent and separable effects on VDI, CDI, and channel activation. The separable effects on VDI are consistent with the previous models of VDI, which include the distal S6 region as a locus for the binding of a “hinged-lid” (Stotz et al., 2000; Stotz and Zamponi, 2001b; Tadross et al., 2010). Notably, the VDI-selective L762F mutation resides distal to the IIS6 region of the channel, which was previously shown to be a critical determinant of the rate of VDI (Stotz and Zamponi, 2001a). More surprising, however, is the selective effect of I1475M on CDI. Prior work has suggested that the impact of S6 mutations on CDI is often secondary to a change in the activation energy required to open the channel (Dick et al., 2010; Tadross et al., 2010). Here, we demonstrate that this is not always the case, and the S6 may participate in transducing CDI more directly. Interestingly, this effect was only seen in the distal IVS6 region, which connects the transmembrane region to the C-tail of the channel, where the Ca2+ sensor CaM resides. As such, it may be that the I1475M mutation disrupts the transduction of CDI following Ca2+ binding to CaM.
While the lack of changes in channel activation is remarkable for I1475M, the opposite is true for I1166T. For this mutation, a dramatic increase in mode 2 gating was apparent in the single channel recordings, resulting in a significant left shift in channel activation and increased open probability across voltages (Fig. 4, G–I). Previous studies have also suggested increased mode 2 gating in the context of a TS mutation. Erxleben et al. (2006) showed an increase in mode 2 gating in CaV1.2 channels harboring the G406R mutation; however, in this case, the entry into mode 2 was thought to be controlled by an aberrant CaMKII phosphorylation site (Erxleben et al., 2006). However, insertion of the I1166T mutation does not form a consensus CaMKII sequence, and insertion of a V at the same site also increased the propensity for mode 2 gating, making it unlikely that this mutation represents a phosphorylopathy. However, it is interesting to note that I1166 resides within a ring of hydrophobic S6 residues that are believed to stabilize the closed state of the channel (Hering et al., 2018). Mutation of the hydrophobic “I” to a neutral “T” residue may play a role in the enhanced opening of channels harboring the I1166T mutation. The far lesser effect of the I1166V mutation may then reflect the lack of a significant change in hydrophobicity but, nonetheless, points to the importance of the I1166 residue in stabilizing the closed configuration as this mutation also produced a left shift and enhanced mode 2 gating, albeit to a far lesser extent. Likewise, the lack of an activation shift for the I1475M mutation, which resides within the IVS6 hydrophobic region, may fit with the preservation of the hydrophobicity of this residue.
It should be noted that whole-cell recordings have previously been carried out on the mutations described in this study (Boczek et al., 2015a; Wemhoner et al., 2015; Landstrom et al., 2016), sometimes with results that are inconsistent with the current study. In particular, E1496K and I1475M were previously reported to produce a left shift in channel activation (Wemhoner et al., 2015). While it is not clear why these results differ, experimental conditions were quite different in that prior experiments were carried out in oocytes and activation curves were determined by a fit to the current–voltage relation. Given the important implications of the lack of activation shift for I1475M, our results from tail current protocols were confirmed by single-channel recordings, demonstrating no activation shift under our experimental conditions.
Overall, this study provides an in-depth analysis of the biophysical mechanisms underlying select CaV1.2 channelopathic mutations. By focusing primarily on mutations within the S6 region, this study advances our understanding of the role of this channel segment in both activation and inactivation. Given the growing number of S6-opathy mutations being identified, such structure–function understanding may provide insight into a variety of channelopathies.
Jeanne M. Nerbonne served as editor.
We thank Debora DiSilvestre and Josiah Owoyemi for their dedicated technical support. We also thank John Hussey and Dr. Sara Codding for insightful comments and discussion.
This project was supported by an National Institutions of Health/National Heart, Lung, and Blood Institute grant (1R01HL149926) and by an American Heart Association postdoctoral fellowship (20POST35211127).
The authors declare no competing financial interests.
Author contributions: M.A. Bamgboye designed and performed experiments, analyzed data, and wrote the manuscript; K.G. Herold performed experiments, analyzed data, and edited the manuscript; D.C.O. Vieira performed experiments and edited the manuscript; M.K. Traficante performed experiments; P.J. Rogers performed experiments; M. Ben-Johny provided resources and critical discussions, and edited the manuscript; and I.E. Dick conceptualized the project, designed and performed experiments, analyzed data, and wrote the manuscript.
This work is part of a special issue on Structure and Function of Ion Channels in Native Cells and Macromolecular Complexes.