All animal procedures were approved by VA Portland Health Care System Institutional Animal Care and Use Committee (IRBNetID: 1635414–4) in accordance with the US Public Health Service policy on Humane Care and Use of Laboratory Animals and the National Institutes of Health Guide for the Care and Use of Laboratory Animals.

Neocortical neurons were isolated from 1 to 2 day old postnatal mouse pups of either sex as described previously (Martiszus et al., 2021; Ritzau-Jost et al., 2021). Animals were decapitated following general anesthesia with isoflurane and cerebral cortices were removed. Cortices were incubated in trypsin and DNase (5 mg/ml and 0.1 mg/ml for 5 min at 34°C) and dissociated with heat polished pipettes. Dissociated cells were maintained in MEM plus 5% FBS on glass coverslips in an incubator (humidified air and 5% CO₂) at 37 °C. Cytosine arabinoside (4 µM) was added 48–72 h after plating to limit glial division. Cells were used, unless otherwise stated, after 7–35 days in culture.

Cells were visualized with an inverted microscope (Leica DM IRB or Olympus IX70). Whole-cell voltage-clamp recordings were made from cultured neocortical neurons using an Axopatch 200B amplifier with 60%–80% series resistance compensation. Current clamp recordings were made using a Heka EPC10 amplifier. The preparation was continously perfused with solution which contained (in mM) 150 NaCl, 4 KCl, 10 HEPES, 10 glucose, 1.1 MgCl₂, 1.1 CaCl₂, pH 7.35 with NaOH. Synaptic transmission was blocked by the addition of (in µM) 10 CNQX, 10 Gabazine, and 50 APV in extracellular bath solution. In current clamp recordings, 2 mM CsCl was added to the extracellular bath solution to reduce contributions of HCN. Voltage-clamp recordings were made using a caesium methanesulfonate intraceullar solution containing (mM) 135 caesium methansulfonate, 1.8 EGTA, 10 HEPES, 4 MgCl₂, 0.2 CaCl₂, 0.3 NaGTP, 4 NaATP, 14 phosphocreatine disodium, pH 7.2 with TEA hydroxide. Current-clamp recordings were made using potassium-gluconate containing intracellular solution containing (mM) 135 Potassium gluconate, 10 HEPES, 4 MgCl₂, 0.3 NaGTP, 4 NaATP, 10 phosphocreatine disodium, pH 7.2 with potassium hydroxide. The electrode resistance in voltage- and current-clamp recordings were usually 2–3 MΩ and 6–8 MΩ respectively. Voltages have been corrected for liquid junction potentials. All experiments were performed at room temperature (21–23°C).

Neocortical cultures were prepared as above but plated in 25 cm² flasks for gene profiling experiments. After 14 days in culture, cells were collected in RLT lysis buffer with 1% 2-mercaptoethanol in 2 ml RB (Qiagen) tubes and stored at −80°C. RNA isolation and microarray assays were performed in the OHSU Gene Profiling Shared Resource. RNA was extracted using the RNeasy Mini Kit (Qiagen) following the manufacturer’s recommended protocol. RNA quality was verified by Bioanalyzer assay (Agilent Technologies). Labeled target cDNA was prepared using the Applied Biosystems WT Plus protocol with an input of 100 ng total RNA. Processed samples were hybridized to a GeneChipTM Clariom S Mouse Array (Affymetrix/Applied Biosystems). Image processing was performed using Affymetrix Command Console (AGCC) v.3.1.1 software and expression analysis was performed using Affymetrix Expression Console software ver.1.4.1.46. The microarray data are available at NCBI GEO (Accession: PRJNA901951; GEO: GSE218028).

Whole cell voltage-clamp recordings were made using an Axopatch 200B Amplifier, filtered at 5 kHz using a Bessel filter, and sampled at 20 kHz during acquisition. Whole cell current-clamp recordings were made using a Heka EPC 10 amplifier, filtered at 2.9 kHz using a Bessel filter and sampled at 20 kHz during acqusition. Series resistance compensation was performed manually prior to acquisition. Analysis was performed using Igor Pro 8 (Wavemetrics, Lake Oswego, OR). Inactivation curves were generated by plotting the normalized peak VGSC current (I_Na(norm.)) versus conditioning voltage (V) which was fit using the Boltzmann function: I N a n o r m . = I R e s + 1 −   I R e s / 1 + exp − z V − V 0.5 24 where I_Res, z, and V_0.5 represent the residual current resistant to inactivation, the apparent valence, and the mid-point of inactivation respectively. The maximum I_Na(norm.) value predicted by the Boltzmann function was used for normalization. In current clamp experiments, input resistance was measured using the steady state voltage deflection elicited by a 70 pA current injection. Action potential properties were obtained by analyzing traces off-line with IgorPro macros. Action potential threshold was measured as the point at which dV/dt reached 20 mV/ms. Action potential amplitude was defined as the voltage difference between threshold and peak. Action potential half-width was defined as the interval between rising and falling phases of the spike at the point halfway between the peak and the holding potential. All data values were reported as mean (± SEM) or median, if not normally distributed. Statistical significance was determined with appropriate parametric or non-parametric tests (GraphPad Prism 8) as described in the figure legends.

Solutions were applied by gravity from a glass capillary (1.2 mm outer diameter) placed 1–2 mm from the neuron under study. Solutions were switched manually using a low dead volume manifold upstream of the glass capillary. CNQX and Gabazine were supplied by Abcam. Creatine Phosphate was supplied by Santa Cruz Biotech. Cinacalcet was supplied by Toronto Research Chemicals. All other reagents were supplied by Sigma-Aldrich.

Most pharmacological modulators of VGSC currents act directly on the ion channel and inhibit the sodium currents responsible for action potential generation. Like many VGSC inhibitors, cinacalcet appears to affect VGSC inactivation but is unusual because it acts indirectly, via a pathway mediated by G-proteins, to inhibit VGSC currents (Mattheisen et al., 2018). In total, we have observed substantial inhibition of VGSC currents in all of the >400 neocortical neurons we have tested. The neocortical cultures contain structurally diverse neurons and support cells (Figure 1A) which express a wide range of VGSC α and β subunits (Table 1). This indicates several VGSC isoforms are sensitive to cinacalcet-mediated inhibition. To more closely test if the pathway utilized by cinacalcet preferentially involves specific VGSC states, we compared the effect of holding potential, on the rate of cinacalcet-induced inhibition of VGSCs. In voltage clamp recordings, VGSC currents were elicited by test pulses to −20 mV (at 0.5 Hz) in cultured neocortical neurons perfused with Tyrode solution (containing 10 µM CNQX, 50 µM APV, and 10 µM Gabazine to block glutamatergic and GABAergic activity). After establishing a stable VGSC current baseline, 5 µM cinacalcet was applied to the neuron which reduced VGSC current amplitude (Figure 1). The time course of inhibition of VGSC currents by cinacalcet was described by a squared exponential function: I t = A e − t / τ 2 + B (1) where I (t), A, B, t, and τ represent VGSC current amplitude during application, initial VGSC current amplitude, final VGSC current amplitude, time, and time constant respectively (Figure 1D). In contrast, directly-acting inhibitors produce a single exponential pattern of inhibition (Bean et al., 1983; Jo and Bean, 2017). The rate of inhibition was higher at depolarized holding potentials as illustrated by the average diary plots and confirmed both by the voltage-dependence of fractional inhibition at 200 s and the time constant of inhibition (Figures 1C–F). Hyperpolarization of the holding potential over the range of -60 to -100 mV decreased fractional block and increased τ substantially (Kruskal–Wallis test, both p < 0.001; Figures 1E, F). VGSC inhibition by cinacalcet was incomplete at −100 mV. The accelerated inhibition of sodium currents by cinacalcet at depolarized holding potentials probably reflects an increased affinity of the unidentified downstream inhibitory molecule to inactivated states of the VGSCs. These data support the proposal that inhibition of VGSC current by cinacalcet is voltage-dependent and occurs via an indirect pathway.

Preferential binding to a specific channel state will shift the dynamic equilibrium altering the relative fraction of VGSCs that occupy each state, as explained by the modulated receptor hypothesis (Hille, 1977; Hille, 1978; Bean et al., 1983). Initially we tested if VGSC recovery from fast inactivation was affected by cinacalcet. The currents were elicited by a test pulse to −20 mV following a family of 100 ms prepulses (−140 to 20 mV in 10 mV increments) and these control currents were compared with those activated following ∼50% or full inhibition by cinacalcet (5 μM; Figure 2A). Cinacalcet shifted the voltage-dependence of fast inactivation in the hyperpolarizing direction, with a mid-point (V_0.5) of −58.3 ± 0.8 mV in control (n = 8), −69.4 ± 2.5 mV after ∼50% inhibition (n = 8), and −91.9 ± 3.6 mV following full inhibition (one-way ANOVA repeated measures, n = 8, p < 0.0001; Figures 2B, C). The hyperpolarizing prepulses facilitated recovery of only ∼10% of the fully inhibited VGSC currents from inhibition by cinacalcet (Figures 2A, B). Cinacalcet also decreased the apparent valence of the inactivation curves from 4.2 ± 0.2 e (control; n = 8), to 2.9 ± 0.4 e following ∼50% inhibition (n = 8), and 1.9 ± 0.1 e following full inhibition (1-W ANOVA RM, n = 8, p < 0.0001; Figure 2D). These data indicate that cinacalcet stabilized the fast-inactivated state of VGSCs.

To further investigate the inhibitory mechanism of cinacalcet, we explored the possibility that X interacts with the slow-inactivated state of the VGSC. A test pulse to −20 mV, delivered after a 100 ms interval at −120 mV, evoked VGSC currents following a series of 5 s prepulses (−140 to +20 mV in 10 mV increments; Figure 3A). The 100 ms interval at −120 mV facilitated recovery from fast inactivation, so that the test pulse permitted the comparison of the effects of the prepulse on VGSC current recovery from a slow-inactivated state. Slow inactivation was complete at −10 mV in control with ∼45% of VGSC currents still available for activation (Figure 3). In the presence of 5 µM cinacalcet, we observed a substantial increase in the fraction of channels recovering slowly (96 ± 0.6%; n = 6) compared to control (56 ± 4%; n = 6; Figure 3B). We also observed a significant increase in the rate of slow inactivation development, with theapparent valences increasing from 2.3 ± 0.2 e in control (n = 6) to 4.5 ± 0.5 e following inhibition by cinacalcet (n = 6, p = 0.004; Figure 3C). In addition, the voltage-dependence of slow inactivation shifted in the hyperpolarizing direction in the presence of cinacalcet, with a midpoint of −59.2 ± 1.7 mV in control and −97.3 ± 1.2 mV (p = 0.004) following inhibition by cinacalcet (Figure 3D). Taken together, these results indicate that cinacalcet apparently stabilizes the slow-inactivated VGSC state.

We attempted to further characterize the kinetics of inhibition by cinacalcet in our neocortical neurons using voltage protocols designed to differentiate between the relative affinities for fast and slow-inactivated VGSC states (Kuo and Bean, 1994; Errington et al., 2008). The presence of channels in the slow-inactivated state was measured with a variable length (0–16 s) prepulse to either −70, −50, −20, or +10 mV, followed by a brief recovery period at −120 mV to allow recovery from fast inactivation, and a test pulse to -20 mV (Figure 4A). Figure 4B shows the induction of slow inactivation at different voltages in the absence of cinacalcet. There was minimal slow inactivation at −70 mV; almost all VGSCs recovered from inactivation during the 50 ms recovery period even after 16 s at −70 mV. The occupancy of the slow-inactivated state increased as the inactivating pulse was made more positive, reaching ∼93% with a 16 s pulse at +10 mV.

Figure 4C shows the development of VGSC current inhibition during application of 5 µM cinacalcet at different depolarized voltages. The inhibition by cinacalcet can be compared directly with the rate and voltage dependence of slow inactivation shown in Figure 4B, because the pulse protocols are identical and the individual experiments were paired. Importantly, however, the 50 ms pulse to −120 mV that eliminates fast inactivation in the control setting may not have the same effect during inhibition by cinacalcet. If fast-inactivated VGSCs become bound during the variable length prepulse, then the 50 ms pulse to −120 mV will hypothetically only recover those fast-inactivated channels that are unbound. Thus, in control, the test pulse assays channels that are in the slow-inactivated state whereas in the presence of cinacalcet, the test pulse will also assay channels that are bound by X.

At each voltage, the kinetics of inhibition by cinacalcet were substantially faster than the development of slow inactivation. At −70 mV, cinacalcet inhibition developed with a fast time constant of 0.03 s and a slow time constant of 1.46 s whereas slow inactivation developed with a fast time constant of 3.46 s and a slow time constant of 30.9 s (Figure 4D). At −50, −20, and +10 mV, cinacalcet inhibition developed with fast time constants of 0.16, 0.23, and 0.20 s respectively, and slow time constants of 2.7, 3.9, and 1.5 s respectively. Conversely, slow inactivation developed with fast time constants of 1.2, 1.7, and 1.4 s at −50, −20, and +10 mV, respectively, and slow time constants of 10.3, 8.5, and 5.7 s. The results with a prepulse to −70 mV demonstrate most clearly the lack of selective binding of X to the slow-inactivated state. There was minimal slow inactivation even with a prepulse to −70 mV for 16 s, yet there was substantial inhibition by cinacalcet (∼55%). These results indicate that cinacalcet does not promote the binding of X exclusively to the slow-inactivated state, as development of inhibition could never be faster than the development of slow inactivation if this were the case.

Previous work has shown that sustained hyperpolarization slowly reverses cinacalcet-mediated inhibition of VGSC currents (Mattheisen et al., 2018). We tested if holding potential (−60, −80, or −100 mV) strongly affected the dynamics of recovery from inhibition. A double-pulse protocol (S₁ and S₂, each to −20 mV for 10 ms) was used to elicit VGSC currents (I₁ and I₂) in control or after complete inhibition by 5 µM cinacalcet (Figure 5A). The ratio I₂/I_1Con reflected the amplitude of the VGSC current elicited by S₂ compared to that elicited by the step to −20 mV from the holding potential before drug application. At −80 mV, I₂/I_1Con recovered fully within 10 ms before cinacalcet application (Figure 5B). Cinacalcet attenuated and slowed the recovery of I₂/I_1Con substantially, so that it eventually reached between 31% and 53% of the control I₂/I_1Con after 8 s at −120 mV (Figure 5B). The majority of this recovery was well described by a single exponential where τ was between 1 and 1.9 s (red curves, Figure 5B). The holding potential had a much greater effect on the dynamics of the gating of the control VGSC currents than the recovery following inhibition by cinacalcet.

Models that describe complex VGSC function indicate the channels can occupy multiple fast and slow inactivation states (Ulbricht, 2005; Milescu et al., 2010; Goldfarb, 2012; Cervenka et al., 2018). The indirect mechanism of inhibition by cinacalcet adds further complexity. To reduce the number of parameters we described the action of cinacalcet using a simpler model with only single fast and slow inactivation states. Two exponential phases of recovery of I₂/I_1Con that represent VGSC repriming (Figures 5B,C, black circles), were observed in the absence of cinacalcet and represent the fast and slow inactive states. The amplitudes of the total I₂/I_1Con value (T, upper panel Figure 5B) and the amplitudes of each exponential were used to estimate the fraction of VGSCs in resting (R, black), fast-inactivated (FI, green), and slow-inactivated (SI, purple) states. The holding potential substantially changed the fraction of VGSCs in the resting state, which corresponded to a I₂/I_1Con value of unity (red broken line, Figure 5B), so that R decreased relative to T as holding potential was depolarized - and, the fraction of VGSCs in the resting state was 1/T. The FI component was defined as any component of I₂/I_1Con below the asymptote for the faster exponential fit (black broken) above the resting component (I₂/I_1Con = 1, broken red line). The SI component was the difference between the asymptote to the double exponential fit (black solid) and the higher of the asymptote of the faster exponential fit (black broken) or I₂/I_1Con = 1. The fraction (F) of the VGSCs in each state was obtained by dividing each component by T (F _R , F _FI , and F _SI ) at each value of V_h. Using the law of mass action and the observation that reversal of VGSC inhibition was relatively slow [Figure 5B and (Mattheisen et al., 2018)], the rate of inhibition of VGSCs (dB/dt) by the unidentified inhibitory molecule was directly proportional to the sum of the products of F and k for each channel state: d B d t   α   X     F R .   k R + F F I .   k F I + F S I .   k S I   where [X], represents the unknown concentration of inhibitory molecule, and k represents the association constants for the three VGSC states. By incorporating a constant, C: d B d t = X   F R .   k R + F F I .   k F I + F S I .   k S I     C (2)

Using Eq. 2, the voltage-dependent rates of inhibition (1/ τ , Figure 1F), and the values of F for each of the states at V_h (Table 2) we constructed a simultaneous equation for each of the three holding potentials. These three equations were solved to return the relative association rates for the R, FI, and SI states of 1, 10.1, and 2.3 respectively.

Cinacalcet binding to an unidentified receptor triggers a pathway resulting in the generation of an inhibitory molecule that preferentially inhibits the fast-inactivated VGSC state. Action potentials are near digital signals in contrast to the graded VGSC current amplitude. The non-linearity of action potentials and the complexity of action of cinacalcet, make it hard to predict how the pathway modulated by cinacalcet will impact action potential generation. We tested how cinacalcet impacted excitability in neocortical neurons at membrane potentials between -60 and -80 mV. Whole-cell current clamp recordings were performed in the presence of (in µM) 10 CNQX, 10 Gabazine, and 50 APV to prevent confounding by the actions of cinacalcet on synaptic transmission. Resting membrane potentials of neurons in this recording configuration were −74.3 ± 1.1 mV (n = 30). Action potentials were elicited by a series of one second current injections (10–70 pA) in neurons current-clamped close to -80 mV (Figure 6A, left panel). Thereafter the recording mode was switched to voltage clamp, to facilitate a consistent holding potential, and 2 µM cinacalcet applied at −80 mV. After 5 min, the recording mode was switched back to current clamp and depolarizing pulses injected from a target holding potential of −80 mV. Cinacalcet was increased to 5 µM and the process repeated as before (Figure 6A, left column). Action potential number decreased at each current injection as cinacalcet concentration was increased and these effects were larger in separate recordings where the membrane potential was targeted to -70 or −60 mV (Figures 6A, B). At each target holding potential, the measured values were well-matched between the control and cinacalcet groups (Table 3). A similar striking effect was observed by comparing the sums of action potentials generated at each holding potential (Figure 6C). Action potential number increased with depolarization of the holding potential and cinacalcet reduced the total number of action potentials generated by the series of current injections.

Using the number of spikes elicited by the 70 pA current injection we examined how the voltage-dependence of cinacalcet inhibition of sodium conductances (Figure 1) impacted action potential generation (Figure 6D). While the average fractional block of action potentials was the same for 2 and 5 µM cinacalcet at −80 mV (42 ± 11% vs. 47 ± 30%), the relative block of the two concentrations increased substantially at −70 mV (38 ± 15% vs. 75 ± 10%). Further depolarization to -60 mV reduced the difference in fractional block between the two concentrations of cinacalcet (53 ± 9% vs. 80 ± 6%). Overall, stronger depolarizations increased the potency of action potential block by cinacalcet.

In addition to reducing the likelihood of action potential generation, VGSC inhibition by cinacalcet may reduce the rate of depolarization and height of the action potential resulting in a shorter and broader waveform. Depolarization may also impact spike waveform by causing VGSC inactivation and changing other voltage-gated channel availability. Input resistances and the properties of the action potentials (the first elicited by a 70 pA depolarizing current) are documented in Table 3 (Mean ± SEM). A two-way ANOVA with repeated measures (2-W ANOVA RM) was performed to analyze the effect of the target holding potential and cinacalcet treatment on input resistance (Table 3) and revealed no interaction (F (4, 54) = 0.696, p = 0.598). Main effects analysis showed holding potential (p < 0.001) but not cinacalcet treatment (p = 0.969) had a significant effect on input resistance. These results suggest cinacalcet did not substantially block ion channels already open at the holding potentials but that conductances were activated by depolarization of the holding potential.

In some recordings, action potentials were not generated following treatment with cinacalcet or depolarization resulting in “missing values”. Consequently, we used mixed-effects analyses to determine the effect of the target holding potential and cinacalcet treatment on the action potential properties. As shown, (Table 3) action potential amplitude and half-width were both affected. Cinacalcet treatment and holding potential interacted to affect action potential amplitude (F (4, 48) = 180.1, p <0.001). Main effects analyses showed that both factors independently affected action potential amplitude (both p < 0.001). There was not a significant interaction of cinacalcet treatment and holding potential on action potential half-width (F (4, 48) = 1.439, p = 0.236). Cinacalcet treatment, but not holding potential, had a significant effect on half-width (p = 0.009 and p = 0.130). The target holding potential and cinacalcet treatment did not interact to affect action potential threshold (F (4, 48) = 1.169, p = 0.336) or affect threshold independently (p = 0.100 and = 0.191, respectively). The smaller, broader action potentials observed at lower frequency following cinacalcet treatment are consistent with VGSC inhibition.