|Home | About | Journals | Submit | Contact Us | Français|
Alternative splicing is known to alter pharmacological sensitivities, kinetics, channel distribution under pathological conditions, and developmental regulation of VGSCs. Mutations that alter channel properties in NaV1.7 have been genetically implicated in patients with bouts of extreme pain classified as inherited erythromelalgia (IEM) or paroxysmal extreme pain disorder (PEPD). Furthermore, patients with IEM or PEPD report differential age onsets. A recent study reported that alternative splicing of NaV1.7 exon 5 affects ramp current properties. Since IEM and PEPD mutations also alter NaV1.7 ramp current properties we speculated that alternative splicing might impact the functional consequences of IEM or PEPD mutations. We compared the effects alternative splicing has on the biophysical properties of NaV1.7 wild-type, IEM (I136V), and PEPD (I1461T) channels. Our major findings demonstrate that although the 5A splice variant of the IEM channel had no functional impact, the 5A splice variant of the PEPD channel significantly hyperpolarized the activation curve, slowed deactivation and closed-state inactivation, shifted the ramp current activation to more hyperpolarized potentials, and increased ramp current amplitude. We hypothesize a D1/S3–S4 charged residue difference between the 5N (Asn) and the 5A (Asp) variants within the coding region of exon 5 may contribute to shifts in channel activation and deactivation. Taken together, the additive effects observed on ramp currents from exon 5 splicing and the PEPD mutation (I1461T) are likely to impact the disease phenotype and may offer insight into how alternative splicing may affect specific intramolecular interactions critical for voltage-dependent gating.
Voltage-gated sodium channels (VGSCs) are dynamic membrane-spanning proteins that play a crucial role in determining the electrical excitability in nerve and muscle.1 Nine isoforms of the highly homologous pore-forming α-subunit (NaV1.1–9, 220–260 kDa) have been identified, each with subtle differences in channel properties likely due differences in amino acid sequence. The well-conserved topology of the VGSC α-subunit structure consists of four domains (D1–D4). Each domain contains six transmembrane α-helical segments (S1–S6) that possess unique properties important for retaining a definitive channel structure. For example, evidence indicates the voltage sensing components are the S1– S4 segments in each domain, where the highly charged S4 segments are the mobile charge translocators, 2–4 and the S5 and S6 segments form the aqueous ion-conducting channel pore.5 Specifically, transitions between ion conducting (opened) and non-conducting (closed and inactivated) gating states have been linked to particular domains and segments of the channel that cooperatively result in changes to the channel configuration, thus allowing or restricting ion movement through the pore.6 Transition between gating states, in part, is determined by a focused electric field7, 8 which supplies voltage-sensitive channels with the potential energy required to surmount conformation-dependent energy barriers.9 The sequence conservation and redundancy between each of the segments in various VGSC isoforms is likely to contribute to their broad similarities (ion selectivity and voltage-dependence), whereas differences in sequence between isoforms may play a role in evolutionary adaptation, such as pharmacological sensitivities (TTX-sensitive vs. TTX-resistant), kinetic properties (fast vs. slow activating/inactivating channels), or regional distributions. As such, single-point missense mutations in “hotspot” regions of the channel, critical for precise gating, can disrupt the ability of the channel to transition between states and alter the configuration required to stabilize the channels at a particular membrane potential.
Several mutations that alter gating properties in NaV1.7, highly expressed in peripheral and sympathetic nervous systems, have been genetically implicated in patients with bouts of extreme pain classified as inherited erythromelalgia (IEM) or paroxysmal extreme pain disorder (PEPD). Remarkably, the locations of the perceived pain are distinct between IEM and PEPD. Additionally, patients with different IEM or PEPD mutations report differential age onsets.10, 11 The contributing factors to these differences are not clear. However, post-translational modification, alternative splicing, and expression of auxiliary β-subunits (β1–4, 32–36 kDa) have been reported to modulate sodium current properties.12 Changes in NaV1.7 biophysical properties resulting from splice variants have been hypothesized to contribute to the differences in perceived location of pain and age-onset between IEM and PEPD. Indeed, alternative VGSC splicing has been shown to alter channel kinetics, pharmacological sensitivities, and tissue distribution under pathological conditions, and can also be developmentally regulated.13–21 Additional evidence suggests that alternative splice variants of NaV1.7 are found in dorsal root ganglia (DRG) neurons and that under modeled neuropathic pain conditions in rats splicing patterns of this isoform can change.22 Furthermore, a recent study23 reported that alternative splice variants of NaV1.7 exon 5 (5N and 5A), which differ by two amino acids (L201V, N206D) in the D1/S3–S4 linker, differentially affect the amplitude of currents generated in response to a slow depolarizing ramp stimulus. IEM and PEPD mutations also alter NaV1.7 ramp current properties and this is thought to contribute to the chronic pain induced by these mutations. Therefore, we investigated whether alternative splicing of exon 5 might impact the functional consequences of IEM (I136V) and PEPD (I1461T) mutations. Results from these experiments (1) suggest an additive effect of the NaV1.7 exon 5 alternative splicing and the PEPD mutation (I1461T) which could further impact the disease phenotype and (2) offer insight into how alternative splicing within exon 5 affects specific intramolecular interactions critical for voltage-dependent gating.
In the present study, hEK293 cells were transiently transfected with α-subunit channels (NaV1.7 wild-type or mutant, 5A or 5N splice variant, cDNA) plus the auxiliary subunits β1 and β2. Electrophysiological properties of the channels were examined using standard whole-cell voltage-clamp techniques.
Wild-type (WT) NaV1.7 exon 5 splice variants (5N and 5A) produced rapid activating and inactivating inward sodium current upon membrane depolarization (Fig. 2A). Initial comparison of the current traces did not suggest noticeable differences (p > 0.05) in channel kinetics or expression levels of the 5N (−374.3 ± 65 pA / pF, n = 13) compared to the 5A (−320.7 ± 44 pA / pF, n = 11) form. Changes in activation (m∞ / V), determined by whole-cell ionic conductance, and inactivation (h∞ / V), determined by the fraction of channels available after a 500 ms conditioning prepulse, were examined for the WT splice variants by comparing their respective midpoints (V1/2) and slope factors (Z) in response to changes in command voltage (Fig. 2B). Additionally, the rapid, integrated transition from channels activating and opening to closing, or deactivation,24, 25 was evaluated (Fig. 2C). In agreement with Chatelier et al.23 a small hyperpolarizing (~2 mV) shift in the V1/2 of activation for the WT 5A variant was observed, whereas changes in steady-state inactivation and deactivation kinetics (τd), at potentials from −100 to −40 mV, were not observed (Table 1). These results suggest the D1/S3–S4 splicing differences between the NaV1.7 variants do not significantly alter steady-state activation, deactivation, or inactivation properties in response to changes in membrane potential.
The relatively slow transitions of NaV1.7 to and from closed-inactivated states are thought to contribute to production of physiologically relevant ramp current in response to slow depolarizing stimuli at hyperpolarized potentials near resting membrane potential for nociceptive DRG neurons.26, 27 These NaV1.7 ramp currents have been proposed to play roles in amplifying subthreshold generator potentials at relatively hyperpolarized voltages.28, 29 Thus, the ramp current generated by NaV1.7 channels is implicated as a critical component in determining action potential firing threshold.30 For these reasons, as well as recent reports suggesting differences in ramp current between NaV1.7 splice variants,23 we examined development of and recovery from a closed-inactivated state (Csi) at −60 mV and changes in the ramp current elicited. Development of Csi (τ−60) for the WT 5N (60.4 ± 6.3 ms, n = 11) compared to the 5A (73.4 ± 12.2 ms, n = 11) form was not significantly (p > 0.05) different (Fig. 2D). Comparison of the recovery from −60 mV did not reveal a significant (p > 0.05) difference between the time constants for WT 5N (10.4 ± 1.1 ms, n = 8) and 5A (9.5 ± 0.7 ms, n = 10). The ramp current generated in response a slow depolarizing ramp stimulus (0.27 mV / ms) by the 5A variant was increased by approximately 40% compared to the 5N form (Fig. 2E). Although this increase was not significantly (p > 0.05) different between the two WT splice variants, it is consistent with that shown previously by Chatelier et al.23 This relatively small increase in ramp current could, in part, be due to the moderate shift in activation and the trend towards a slowed development of Csi that was observed compared to the WT 5A splice variant.
We next examined the effects alternative splicing has on a PEPD mutation (I1461T) located within the putative D3–D4 intracellular inactivation gate (Fig. 1). The I1461T mutation, and other PEPD mutations that have been previously characterized in the 5N splice variant, shift the voltage-dependence of steady-state inactivation in the positive direction, slow the rate of open-state fast-inactivation and increase the amplitude of ramp currents at positive potentials compared to WT 5N channels.31 The whole-cell inward sodium current generated by the I1461T 5N and 5A splice variants were similar (Fig. 3A) and current expression was not statistically (p > 0.05) different between the 5N (−171.2 ± 26 pA / pF, n = 12) and the 5A (−179.2 ± 27 pA / pF, n = 12) variants. Steady-state inactivation properties (h∞ / V) were not significantly (p > 0.05) different, however the ~5 mV hyperpolarizing shift in the midpoint of activation (m∞ / V) for I1461T 5A was significantly (p < 0.05) different compared to the 5N form (Fig. 3B, Table 1). Furthermore, the deactivation time constants (τd) for the I1461T 5A form were slowed compared to the 5N. For example, deactivation at −60 mV, a potential near resting membrane potential for nociceptive neurons, was significantly (p < 0.05) slowed for the 5A splice variant (Fig. 3C). Examination of Csi at −60 mV also revealed the I1461T 5A channels transition to a close-inactivated state with a significantly (p < 0.05) slower time constant (τ−60) than the 5N channels (Fig. 3D). However, recovery from Csi was not significantly (p > 0.05) different between the 5N (4.7 ± 0.3 ms, n = 11) and the 5A (4.9 ± 0.2 ms, n = 11) variants. Taken together, these data indicate that, at −60 mV, the 5A variant of the I1461T channels have a higher probability of remaining in an activated state, and may be more likely to transition to an ion-conducting state at more hyperpolarized voltages compared to the I1461T 5N variant. When the 5A PEPD mutant ramp current was compared to the 5N channels an increase in peak ramp current elicited was observed at all potentials tested. Furthermore, the 5A PEPD mutant channels displayed a more hyperpolarized onset of ramp current activation (Fig. 3E). These results demonstrate that the I1461T 5A splice variants significantly increases ramp current amplitude at negative and positive potentials compared to the 5N variant.
Mutations associated with IEM are commonly located in D1 and D2 and predominantly shift the voltage-dependence of activation to more hyperpolarized potentials without disturbing steady-state inactivation.32 We examined the effects of alternative splicing on a nearby D1/S1 (Fig. 1) IEM mutation (I136V), which exhibits a delayed age of onset for pain sensations compared to other IEM mutations.33 The currents generated by the 5N and 5A splice variants of the I136V mutant were similar (Fig. 4A) with no statistical (p > 0.05) difference in current expression between the 5N (−339.1 ± 85 pA / pF, n = 6) and the 5A (−214.9 ± 31 pA / pF, n = 9) variants. Further inspection of the voltage-dependent channel characteristics did not yield significant (p > 0.05) differences between the I136V splice variants for the ability to activate during the m∞ / V protocol, inactivate during the h∞ / V protocol, or deactivate (τd) at the range of transmembrane potentials measured (Figs 4B–C). Additionally, no changes in the ability of the I136V-IEM splice variants to transition to a closed-inactivated state were observed at −60 mV (Fig. 4D). Recovery from Csi between the 5N (10.5 ± 1.0 ms, n = 3) and the 5A (10.5 ± 2 ms, n = 3) IEM variants also did not differ (p > 0.05). The ramp current elicited for the D1/S1 IEM mutant was not statistically (p > 0.05) different when comparing the percentage of peak current between the splice variants (Fig. 4E). It is interesting to note that, in contrast to WT and I1461T channels, there was not even a hint of an increase in ramp current amplitude. These results indicate that differences in the splice variants do not affect the I136V mutant channel properties.
In this study we present evidence that the 5A alternative splice variant of NaV1.7 alters the biophysical properties of a disease mutation implicated in PEPD (I1461T) but not IEM (I136V). Since a majority of the IEM mutations have been shown to primarily alter the voltage-dependent activation and deactivation properties (for a review see ref. 27) the relatively small shifts induced by the 5A splice variant on voltage-dependent activation may be masked. The I136V mutation alone shifted the midpoint of activation by approximately −8 mV compared to WT, which is much larger than the shift observed for the WT-5A compared to WT-5N splice variants. Our results suggest that IEM mutations are not additively affected by the 5A splicing. These results are in agreement with those from Han et al.34 showing that the voltage-dependent properties of another “late-onset” IEM mutation (Q10R) are also not affected by alternative splicing. However, our results for the PEPD (I1461T) splice variants are intriguing because they indicate that the impact of PEPD mutations and exon 5 changes are additive. Mutations associated with PEPD generally affect the voltage-dependent transition to an inactivated state and thus, destabilize an inactivated channel configuration allowing an increasing fraction of channels to remain available for activation and opening.11, 31, 35 Many of the characterized PEPD mutations shift the voltage-dependent inactivation properties to more depolarized potentials and yield incomplete development of closed-inactivation and accelerated recovery from inactivation compared to wild-type.31 Conversely, our results demonstrate that the 5A splice variant of WT channels moderately hyperpolarizes the midpoint of activation and increases ramp current, which are in agreement with Chatelier et al.23 Since the stability of the inactivated conformation is decreased by the I1461T mutation, the effects of the 5A alternative splicing for the I1461T channels would be predicted to amplify the overlap between channels opening versus inactivating (window current) and increase the probability of channel opening in an additive manner. Indeed, when the ability of channels to generate subthreshold ramp currents was assayed the I1461T 5A variant significantly increased the peak current and shifted the ramp current threshold to more hyperpolarized potentials compared to the 5N variant. The changes observed in ramp current for the I1461T 5A variant could be caused by a lowered energy barrier for the channels to activate relative to deactivation, and a slowed and incomplete development of a closed-inactivated state thus favoring ion flux (Fig. 6). It is worthwhile to note that the voltage-dependent ramp current properties of the I1461T 5A variant, even when compared to the wild-type 5A variant, resemble a melding of those observed for IEM mutations33, 36, 37 and PEPD mutations31, 35 in NaV1.7 channels similar to those observed for a novel mutation that caused patient symptoms associated with both disorders.38 This feature is particularly interesting because patients diagnosed with PEPD have reported differential age onset of pain compared to IEM patients and increased incidence of pain later in life.39 Therefore, it is appealing to consider that temporal and tissue specific increases in relative expression of the 5A variant could result in a transition to a state where patients with disease mutations in NaV1.7 experience more frequent and diverse attacks.
The coding region for NaV1.7 exon 5 includes a portion of the D1/S3 segment through the C-terminal end of the positively charged D1/S4.22 The residue differences between the 5N and the 5A variants are localized to the extracellular portion of the D1/S3 and the D1/S3–S4 cytosolic linker. Interestingly, the residue difference in the cytosolic linker of the 5A variant contains a polar, negatively charged (acidic) residue (Asp206) compared to the 5N variant which retains a polar, non-charged Asn at the 206 position at a physiological pH (Fig. 5B). This acidic residue substitution could contribute an additional negative charge to the electrostatic field surrounding the D1/S4.7, 8 Additionally, the Asp206 may form electrostatic interactions with N-terminal residues in the mobile D1/S4 charge translocator, which is laden with positive residues.40, 41 In fact, alternative splicing in this region of NaV1.5, where a positively charged residue (Lys) has been substituted for a negative residue (Asp) between variants, has been explored by Onkal et al.42 Their results suggest that the charge reversal impacts the voltage-sensitivity and kinetics of channel activation, such that the positively charged splicing substitution within D1/S3–S4 depolarized the activation midpoint and slope factor, and slowed the time to peak, all of which were reversed upon reversing the charged residue substitution. Since movement of the D1/S4 is critical during the early fast component of gating charge translocation and channel activation,43, 44 a negative charge on an extracellular linker within proximity to accessible positively charged D1/S4 residues could accelerate the movement of this segment during depolarization. Indeed, several studies have shown that the upper portion of charged residues within the S4 is accessible in the extracellular space during depolarization.45–48 Thus, small changes such as increased electrostatic interactions (Fig. 5B) that may lower the energy barrier for activation for the S4 segments, would be predicted to enhance their gating movement, reduce deactivation, and increase the probability to generate subthreshold currents at more hyperpolarized potentials. However, although the other amino acid change that results from the alternative splicing involves a more conserved substitution (Leu to Val at position 201), conserved substitutions can have important consequences (i.e. I136V causes chronic pain) and therefore, based on our data, we cannot rule out a role for this amino acid change in altered channel gating. Furthermore, we also cannot overlook the possibility that the combination of the two substituted residues that differ in the 5N versus the 5A variants (Fig. 5B) decreases the stability of the closed state of the channel population relative to the open state. And finally, it is worthwhile to consider the possibility that particular VGSC auxiliary subunits, expressed in nociceptive neurons, may augment channel properties in response to regulation of alternative splicing events.
Alternative splicing has been shown to affect the properties of a variety of sodium, 13–16, 22, 42, 49 calcium, 50–56 and potassium channels.57–59 In particular, NaV1.1–3, 1.5 and 1.6, in addition to NaV1.7, have been shown to have alternative splicing within exon 5 that involves a change in the charge of one of the residues.22 Interestingly, of the NaV isoforms with exon 5 splicing, disease-associated (gain-of-function) mutations have also been identified in NaV1.1 and 1.2 (epilepsy),60 and 1.5 (LQTS)61 and, as with the PEPD mutations, some of these mutations primarily impair inactivation. Therefore, based on our data with the I1461T PEPD mutation, we predict that the changes induced by splicing are likely have a functional impact on other disease mutations and this could be useful in understanding the underlying molecular mechanisms. It is likely that expression patterns of splice variants in human populations are altered during painful conditions16, 22, 56 and may result in unique differences in response to pharmacological agents.62 Although the interactions and effects may be complex, physiological regulation of the splice variants may prove useful in better understanding disparities in pathophysiology related to patient age or genetic background. Understanding how alternative splicing alters intramolecular interactions during channel gating and impacts the consequences of disease mutations could lead to the development of novel therapeutic strategies.
Human embryonic kidney (hEK293) cells were obtained from ATCC, Manassas, VA, USA. Use of the hEK293 cells was approved by the Institutional Biosafety Committee and conformed to the ethical guidelines for the National Institutes of Health for the use of human-derived cell lines.
Human embryonic kidney (hEK293) cells were transiently transfected with the human NaV1.7 (hNaV1.7 wild-type (WT) and mutant (Δ) I136V and I1461T) splice variant α-subunit cDNA and human β1 (hβ1) and β2 auxiliary subunits (hβ2)63 utilizing the calcium phosphate precipitation technique. Alternative splicing of the SCN9A gene, which encodes for the human NaV1.7 channel, occur within exon 5 which encodes part of the D1/S3 and all of D1/S464 (refer to Figs. 1 and and5B).5B). For the present studies, we performed experiments using the 5N11RS (Leu201 and Asn206, abbreviated as 5N,) and 5A11RS (Val201 and Asp206, abbreviated as 5A) short form splice variants which differ by two amino acids in the D1/S3–S4 cytoplasmic linker at positions 201 and 206 (Figs. 1 and and5B5B).22 Mutagenic primers were designed to introduce the correct base pair change into the hNav1.7 channel using Vector NTI Advance 10 (Invitrogen, Carlsbad, CA, USA). Mutations inserted into the hNav1.7 channel cDNA65 construct were produced using the QuickChange XL Site-Directed Mutagenesis Kit (Stratagene, La Jolla, CA, USA) according to the manufacturer’s protocol. hEK293 cells were grown under standard tissue culture conditions (5% CO2; 37°C) in Dulbecco’s modified Eagle’s medium (DMEM) supplemented with 10% fetal bovine serum (FBS). The calcium phosphate-DNA mixture was added to the cell culture medium and left for 3–4 hours, after which the cells were washed with fresh medium. Sodium currents were recorded 48–72 hours post-transfection.
Fire-polished electrodes (1.0–2.0 MΩ) were fabricated from 1.7 mm capillary glass using a Sutter P-97 puller (Novato, CA, USA). The standard CsF dominant electrode (intracellular) solution (at 293 mOsm) consisted of (in mM) 140 CsF, 10 NaCl, 1.1 EGTA, and 10 HEPES, pH 7.3 (adjusted with NaOH). The standard bathing (extracellular) solution (at 304 mOsm) consisted of (in mM) 140 NaCl, 1 MgCl2, 3 KCl, 1 CaCl2, and 10 HEPES, pH 7.3 (adjusted with NaOH). Cells on laminin-coated glass coverslips were transferred to a recording chamber containing 250 μL of bathing solution. Whole-cell patch-clamp recordings were conducted at room temperature (~22.0 °C) using a temperature gauge (Fisher Scientific, USA) after obtaining a Giga-ohm seal (1–20 GΩ) using a HEKA EPC-10 amplifier under voltage-clamp mode. Cells were selected based on their ability to express EGFP. Offset potential was zeroed before patching. Capacitive artifacts were cancelled using the computer-controlled circuitry of the patch clamp amplifier. Series resistance errors were always compensated to be less than 3 mV during voltage-clamp recordings and leak currents were linearly cancelled by digital P/-5 subtraction. Offset potential was zeroed before patching. Cells were held at a membrane potential of −120 mV. Membrane currents were filtered at 5 kHz sampled at 20 kHz and tail currents were filtered at 10 kHz and sampled at 40 kHz. Data were not collected before five minutes after whole-cell configuration had been established to allow adequate time for the electrode solution to equilibrate. Whole-cell patch recordings did not last more than 45 min and cells were not held in the standard bathing solution for more than one hour. All inward sodium currents had a reversal potential of ~ +65 mV, corresponding closely to the calculated Nernst potential, observed during the standard I / V protocol. Data were acquired on a Windows-based Pentium IV computer using the Pulse program (v 8.65, HEKA Electronik, Germany).
Holding potential for all recordings was set to −120 mV. Current-voltage (I / V) relationships were determined by an incremental depolarizing step protocol, testing every + 5 mV for 100 ms, from −80 to +60 mV. To determine the fraction of channels transitioning to a fast-inactivated state a double-pulse protocol (h∞ / V) was employed which incrementally conditioned the channels from −150 to −10 mV for 500 ms before testing for the fraction of channels available at 0 mV. Voltage-dependent deactivation kinetics were assessed by eliciting tail currents at a range of potentials after briefly activating the channels (0 mV, 0.5 ms). The averaged voltage-plot was compiled using time constants (τd) obtained from tail current recordings which were determined by fitting each decay component with a single-exponential equation. Time course for the development of closed-state inactivation (Csi) at −60 mV was obtained using a time-varied conditioning pulse at −60 mV for 500 ms before testing for the available fraction of channels at 0 mV for 20 ms. Time constants (τ−60) were determined using a single-exponential fit. Time course for the recovery from closed-state inactivation at −60 mV was tested using a two-step protocol where the channels were conditioned at −60 mV for 500 ms, then hyperpolarized to −120 mV for various times before testing for the fraction of channels available during a depolarizing step to 0 mV for 20 ms. Channel ramp current generation was assayed using a slow depolarizing ramp (0.27 mV/ms) stimulus from −120 mV to +40 mV at a holding potential of −120 mV. Inward ramp current displayed is a result of dividing the individual traces by the peak transient sodium current obtained during the I / V protocol, thus yielding the % of peak current for each recording. Voltage-clamp experimental data were analyzed using the Pulsefit (v 8.65, HEKA Electronik, Germany), Origin (v 7.0, OriginLab Corp., Northhampton, MA, USA), Prism (Prism v 4.0, GraphPad Software, La Jolla, CA, USA) and Microsoft Excel software programs. Normalized conductance-voltage (G–V) relationships were derived using Eq. 1:
Where GNa is macroscopic sodium conductance, Imax is calculated as peak current in response to the test pulse, Vm is the test pulse voltage, and ENa is the measured Na+ equilibrium potential. Normalized availability curves were fit using a standard single-phase Boltzmann distribution for G–V, during the m∞ / V protocol and steady-state fast-inactivation (h∞ / V) data. Midpoint (V1/2) and slope factors (Z) of (activation) conductance-voltage (G–V) and voltage-dependent steady-state fast-inactivation curves were calculated using a standard single-phase Boltzmann distribution fit according to Eq. 2:
All data are shown as means ± S.E.M. Statistical significance was assessed with Microsoft Excel using the Student’s unpaired t-tests. Statistical significance of difference was accepted at p < 0.05.
This work was supported by the National Institutes of Health Research Grant NS053422 (TRC).