Two pairs of CACNA1I (CaV3.3) variants with opposite effects on channel function cause neurodevelopmental disorders of varying severity

Abstract

The T-type voltage-gated calcium channel CaV3.3 is expressed in GABAergic neurons of the thalamic reticular nucleus (TRN), where its pacemaking activity controls sleep spindle rhythmogenesis during the non-rapid eye movement (NREM) phase of natural sleep. Previously, we established CACNA1I, the gene coding for CaV3.3, as a disease gene for neurodevelopmental disease with or without epilepsy. Here we report three newly identified activation-gate-modifying heterozygous missense variants of CACNA1I, found in four unrelated patients with neurodevelopmental disease with or without seizures. One of these variants, p.(Met1425Val), is an amino-acid substitution at the same position as previously published variant p.(Met1425Ile). Notably, the other two variants studied here are also a pair of two different substitutions of the same amino acid: p.(Ala398Val) and p.(Ala398Glu). By using site-directed mutagenesis, voltage-clamp electrophysiology, computational modelling of neuronal excitability, and structure modelling, we found that the two substitutions of M1425 both result in a gain of channel function including left-shifted voltage-dependence of activation and inactivation, slowed inactivation and deactivation kinetics, and increased neuronal excitability. Remarkably, the two substitutions of A398 show opposite effects on channel function. While substitution A398E leads to a gain of channel function, A398V results in decreased current density, accelerated gating kinetics, and a decreased neuronal excitability. The lack of seizures in the two independent p.(Ala398Val) patients correlates with the absence of increased neuronal excitability in this variant. This is the first report of a gate-modifying CaV3.3 channel variant with partial loss-of-function effects associated with developmental delay and intellectual disability without seizures. Our study corroborates the role of CaV3.3 dysfunction in the etiology of neurodevelopmental disorders. Moreover, our data suggest that substantial gain-of-function of CaV3.3 leads to the development of seizures, whereas both gain- and loss-of-function variants of CACNA1I can cause neurodevelopmental disease.

Fig 1. Location of disease-associated CACNA1I amino acid substitutions in the domain structure and structure modelling of Ca_V3.3.

Voltage sensor domain (VSD) helices S1-S4 displayed in grey, connecting S4-S5 linker in beige, and pore domain (PD) helices S5 in cyan and S6 in purple. (A) Domain structure of Ca_V3.3 indicating the location of the disease-associated variants at the cytoplasmic end of S6 helices. A398E (green) and A398V (red) in IS6, M1425V (magenta) and M1425I (orange) in IIIS6. Indicated in grey are also the three previously published CACNA1I disease-associated amino acid substitutions I860N and I860M in IIS6, and I1306T in IIIS5 (El Ghaleb et al., 2021) [9]. (B) Top-, bottom-, and side view of the structure model of Ca_V3.3 α1 subunit based on the cryo-EM structure of Ca_V3.3 (PDB accession code: 7WLI) in the inactivated state with a closed channel gate. The inset shows a zoom-in with the two changed residues A398 and M1425 displayed.

Fig 2. Distinct effects on calcium channel activation of CACNA1I disease-associated amino acid substitutions A398E/V and M1425V/I.

(A-E) Representative calcium current recordings in response to 500 ms depolarization to voltages between -90 mV to +40 mV in 10 mV increments for (A) Ca_V3.3 wild-type (WT) in blue, (B) A398E (AE) in green, (C) A398V (AV) in red, (D) M1425V (MV) in magenta, (E) and M1425I (MI) in orange. (F, J) The current-voltage relationship of the two A398 variants (F) and the two M1425 variants (J), each set compared to matched WT controls. (G, K) Peak current densities are not altered in AE (p = 0.8791), MV (p = 0.4624), and MI (p = 0.1475), except in AV, where they are reduced by more than 2-fold (p = 0.0348). (H, L) Fractional activation curves and (I, M) V_1/2 of activation scatterplots show significantly left-shifted voltage-dependence of activation for Ca_V3.3 AE by 10.7 mV, MV by 12.3 mV, and MI by 11.0 mV. The V_1/2 of AV is right-shifted by 4.0 mV, all as compared to the respective wild-type controls. Mean ± SEM; p-values calculated with one-way ANOVA and Dunnett’s multiple comparisons test; * p < 0.05, ** p < 0.01, **** p < 0.0001.

Fig 3. Activation, inactivation, and deactivation kinetics differ between Ca_V3.3 WT and variants as well as among the variants.

(A, D) Time constants of activation calculated from fits of the rising phase during 500 ms step depolarisations to the indicated voltages. (A) Both A398E (green) and A398V (red) activate faster between -40 and -20 mV, as compared to the WT (blue). (D) At -30 mV and higher depolarisations both substitutions of M1425 activate as fast as WT. (B and E) Time constants of inactivation determined by fitting the decay phase of currents during 5 s depolarisations. (B) AE inactivates slower and AV faster, compared to WT. (E) Both M1425V (magenta) and M1425I (orange) inactivate slower compared to WT. (C and F) Time constants of deactivation determined from tail current decay at indicated repolarising voltages after 15 ms pulse to V_max. Insets display representative normalized example traces. (C) A398E deactivates increasingly slower, A398V faster only at -60 mV. (F) M1425V and M1425I deactivate increasingly faster compared to WT. Mean ± SEM; p-values calculated with repeated measures ANOVA and the Holm-Sidak test for multiple comparisons. *p < 0.05, **p < 0.01, ****p < 0.0001.

Fig 4. All Ca_V3.3 variants left-shift the voltage dependence of inactivation.

Inset in A shows the steady-state inactivation protocol used for these experiments. (A) The fractional inactivation curves (B) and scatterplot of the V_1/2 of the voltage dependence of inactivation show left shifts for each of the four variants: 10.1 mV for A398E (AE, green), 13.0 mV for A398V (AV, red), 9.1 mV for M1425V (MV, magenta), and 10.3 mV for M1425I (MI, orange). (C) Superimposing fractional inactivation and activation curves (WT activation curve from pooled WT data S1C Fig) demonstrates left-shifted activation and inactivation together result in left-shifted window-currents for AE (green), MV (magenta), and MI (orange), compared to the respective WT control (blue). For AV (red), only the inactivation curve is left-shifted and the activation curve right-shifted, resulting in a reduced window-current voltage-range. Mean ± SEM; p-values calculated with one-way ANOVA and Dunnett’s multiple comparisons test; **** p < 0.0001.

Fig 5. Three Ca_V3.3 variants left-shift the window current, while the A398V completely abolishes the window current.

(A) Representative example traces of a 10 sec. ramp recording from -100 mV to 0 mV of Ca_V3.3 wild-type (WT, blue), A398E (AE, green), A398V (AV, red), M1425V (MV, magenta), and M1425I (MI, orange). Note that the ramp voltage (gray) increases parallel to the time. (B) While the size of the window current, normalized against the peak current density (I window/ I max) is unchanged for AE and not significantly increased in the MV variant, the MI variant shows a significant 2-fold increase of the relative window current. The window current of the AV variant is completely absent. (C) As compared to WT controls, the three variants AE (7.6 mV shift), MV (8.3 mV shift), and MI (6.3 mV shift) all show significantly left-shifted peaks of their window current. Mean ± SEM; p-values calculated with one-way ANOVA and Dunnett’s multiple comparisons test; *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001.

Fig 6. NEURON computer modelling predicts increased electrical activity at rest and in response to hyperpolarizing pulses for three variants of Ca_V3.3 and quiescence at rest for the A398V variant.

(A, C, E, and G) Show the spontaneous firing at rest of each variant compared to WT for 1500 ms. At 1500 ms, a 200 ms hyperpolarizing pulse (-0.088 nA) to -90 mV was injected to test rebound firing. (B, D, F, and H) Show zoom-ins of 350 ms right after the hyperpolarizing pulse. (A, E, and G) A398E (AE, green), M1425V (MV, magenta), and M1425I (MI, orange) show a two Hz increased firing frequency at rest (9 Hz) as compared to WT (7 Hz). (C) A398V (AV, red) does not fire action potentials at rest. (B, D, F, and H) Both wildtype and variants show an increased firing rate after the hyperpolarization to -90 mV; AE, MV, and MI fire with a higher frequency than WT. AV only fires a single action potential.

Fig 7. Three variants lower the rheobase and increase firing frequency of Ca_V3.3, while the A398V variant shows the opposite effect.

The resting membrane potential was set to -90 mV and 200 ms depolarizing current pulses of increasing amplitudes were applied. (A) A lower current injection necessary to stimulate an action potential (rheobase) was observed for A398E (AE, green, 0.051 nA), M1425V (MV, magenta, 0.051 nA), and M1425I (MI, orange, 0.050 nA) and all three fire with a higher frequency at every pulse, as compared to wild-type (WT, blue, 0.065 nA). In the model, A398V (AV, red, 0.073 nA) has lower firing frequencies and higher rheobase than WT. (B-E) Electrical activity at the rheobase. AE, MV, and MI show an increased latency to the start of action potentials compared to WT, while AV shows a faster rising phase and shortened latency than WT.

Fig 8. Structure model predictions of wildtype and A398E/V and M1425V/I Ca_V3.3 channels.

Shown is the comparison between the wild-type structure model and the four variants in the open pore and closed pore states. (A) A398 points into the gate aperture. The clouds, blue in A398 (WT) and red in A398V, show the hydrophobic surface. The side chain of V398 (red) forms a hydrophobic plug (red cloud) together with other hydrophobic residues in IS6 and neighbouring S6 helices, sealing the gate even in its open configuration. E398 (green) forms hydrogen bonds with residues in IS6 (S402) and neighbouring S6 helices, as well as a water network (blue mesh) inside the gate. (B) The two variants of M1425 can both form stabilizing hydrophobic interactions with neighbouring hydrophobic residues pointing away from the gate aperture.