The de novo CACNA1A pathogenic variant Y1384C associated with hemiplegic migraine, early onset cerebellar atrophy and developmental delay leads to a loss of Cav2.1 channel function

CACNA1A pathogenic variants have been linked to several neurological disorders including familial hemiplegic migraine and cerebellar conditions. More recently, de novo variants have been associated with severe early onset developmental encephalopathies. CACNA1A is highly expressed in the central nervous system and encodes the pore-forming CaVα1 subunit of P/Q-type (Cav2.1) calcium channels. We have previously identified a patient with a de novo missense mutation in CACNA1A (p.Y1384C), characterized by hemiplegic migraine, cerebellar atrophy and developmental delay. The mutation is located at the transmembrane S5 segment of the third domain. Functional analysis in two predominant splice variants of the neuronal Cav2.1 channel showed a significant loss of function in current density and changes in gating properties. Moreover, Y1384 variants exhibit differential splice variant-specific effects on recovery from inactivation. Finally, structural analysis revealed structural damage caused by the tyrosine substitution and changes in electrostatic potentials.


Introduction
Voltage-gated Ca 2+ (Ca V ) 2.1 channels are the most abundant Ca V channel in the mammalian brain where they are expressed in all brain structures with particularly high expression in the cerebellum [1,2]. Ca V channels are formed by the pore-forming Ca V α 1 subunit and ancillary Ca V α2δ and Ca V β subunits. Ca V α 1 encompasses 24 transmembrane α-helical segments divided into 4 domains (I-IV), each one with six segments (S1-S6). expression in human cerebellum [11] and 79% in human cortex [12]. Due to their central role in neurotransmission, mutations in Cav2.1 channels are expected to impact synaptic transmission. Moreover, there are several reports where mutations in the Ca V 2.1 gene (CACNA1A) cause several autosomal-dominant neurological disorders including familial hemiplegic migraine type 1, and cerebellar pathologies such as ataxia, progressive ataxia and early-onset cerebellar syndrome [13,14]. Moreover, Ca V 2.1 mutations have been associated with congenital ataxia, characterized by chronic cerebellar syndromes and acute symptoms of either episodic ataxia or hemiplegic migraine [13,15]. Hemiplegic migraine (HM) is a rare type of migraine with aura associated with a transient motor weakness or hemiparesis [16]. HM can be familial (with an autosomal-dominant inheritance pattern) or sporadic (de novo mutations). Here, we describe the functional consequences of a de novo missense mutation c.4151A > G (p.Y1384C) in Cav2.1 channel activity. This mutation is located in the highly conserved transmembrane S5 of the third domain of the Ca V 2.1α 1 subunit (Fig. 1). The Y1384C variant was found in a patient with sporadic hemiplegic migraine, cerebellar atrophy and developmental disability, a phenotype previously reported in brief detail in this individual, as well as in another patient in the literature [17,18]. Our functional studies reveal that this mutation causes a loss of function in the channel.

Patient and sequencing
The patient was ascertained and phenotyped through the clinical practice of one of the authors (AMI). The genetic sequencing was performed as previously reported [18].

Electrophysiology
Electrophysiological recordings were performed at room temperature using whole cell configuration patch clamp with an Axopatch 200B amplifier (Molecular Devices). The external solution consisted of (in mM): 2 CaCl 2 , 137 CsCl, 1 MgCl 2 , 10 HEPES, 10 glucose (pH 7.4 adjusted with CsOH). The pipette solution contained (in mM): 130 CsCl, 2.5 MgCl 2 , 10 HEPES, 10 EGTA, 3 ATP-Mg, 0.5 GTP-Na (pH 7.4 adjusted with CsOH). Data acquisition was performed using pClamp11.03 software. Leak and capacitance components were subtracted on-line using a P/4 protocol. Currents were filtered at 5 kHz. Recordings were analyzed using Clampfit 11.03 and figures, fittings, and statistics (ANOVA) were made using GraphPad Prism 8.0. To ensure accurate comparisons, electrophysiological recordings alternated within the same day for all channel types.

Recording protocols and data analysis
Ca V 2.1 Ca 2+ currents were recorded by applying 250 ms pulses ranging from − 60 to + 25 mV in 5 mV increments from a holding potential (hp) of − 100 mV. Current-density voltage relationships were obtained from the peak current divided by cell capacitance as described previously [19]. The I-V relationships were fitted with a modified Boltzmann equation: I = Gmax × (Vm − Vr)/ (1 + exp(− (Vm − V 1/2 )/k)), where I is the peak current, Vm is the membrane potential, V 1/2 is the voltage for half activation, Vr is the reversal potential, and k is the slope factor. Activation curves were obtained by calculating conductance from I to V curves and plotting the normalized conductance (G/Gmax) as a function of membrane potential. Percentage (%) of inactivation was calculated as the percentage of current inactivated at 250 ms with respect to the peak current amplitude. Steady-state inactivation curves were obtained using depolarizations to − 5 mV for 140 ms following 5 s prepulses from − 100 to 0 mV at 10 mV increments from a holding potentiual of − 120 mV. Curves were constructed by plotting the normalized current (I/Imax) as a function of the prepulse potential and fitted with the equation I/Imax = 1/ (1 + exp(− (Vm − V 1/2 )/k)), where V 1/2 is the voltage for half inactivation, and k is the slope factor. Recovery from inactivation was determined using a two pulse protocol.
The first pulse (2 s) and second pulse (50 ms) were at 0 mV and separated by a varying interval ranging from 20 ms to 7.5 s. Traces were normalized to the maximum current during the first pulse for each sweep and plotted against time. Curves were fitted with a single exponential function. Window currents were plotted by using the values obtained from the fits of conductance and steadystate inactivation curves.

Surface biotinylation
Surface biotinylation was performed as described previously [20]. Briefly, transfected cells were incubated on ice with HEPES-based saline solution (HBSS) for 15 min to stop protein trafficking. Surface proteins were then biotinylated for 1 h with 1 mg/ml of EZLink Sulfo-NHS-SS-Biotin (Thermo Scientific) in HBSS. The reaction was quenched with a solution of 100 mM Glycine in HBSS and cells were washed and lysed in a modified RIPA buffer (in millimolar: 50 Tris, 150 NaCl, 5 EDTA, 1% Triton X-100, 1% NP-40, 0.2% SDS, pH 7.4) for 45 min. Two mg of biotinylated proteins were purified using 100 µl of Neutravidin beads (Thermo Scientific) for 1.5 h at 4 °C. Biotinylated fractions and whole cell lysates were resolved by SDS-PAGE and analyzed by western blot using an anti-Ca V 2.1α 1 antibody (ACC-001, Alomone, 1:500) and anti-Na/K ATPase (1:5000, Abcam AB7671). Densitometric analyses were carried out using the ImageJ program (National Institutes of Health).

Protein structural modeling
Homology models of full length Ca V 2.1α 1 subunits were generated using Phyre 2 [21] using Cav1.1α 1 as a template where 66% of the residues were modelled with > 90% of confidence. Missense mutation analysis was performed using Missense3D [22]. Models of the wild type and mutated of transmembrane segments S5 and S6 as well as the S5-S6 linker domain III were obtained using Phyre 2 and electrostatic potentials were applied based on the full nonlinear solution of the Poisson-Boltzmann equation (23) using the Swiss-PdbViewer software.

Statistical analysis
All error bars reflect standard errors. One-way analysis of variance (ANOVA) and Tukey's multiple comparison test with a single pooled variance test was performed for multiple comparisons. Significance was set at 0.05. Significance was established as follows: *p < 0.05, **p < 0.01, *** p < 0.001, **** p < 0.0001.

Clinical characteristics
The patient is a now 31 year old male. Some basic elements of his history were previously reported ( [18], patient 18). In brief, he was born at 37 weeks gestational age to a 32 year old G1P0 mother. Apgar scores were 5 and 8. Hypotonia was present from the newborn period and he first presented to attention in Clinical genetics at nearly 3 years of age for evaluation of his developmental delay and hypotonia. Early computerized tomography (CT) of the brain was reported as normal. Subsequent magnetic resonance imaging (MRI) scanning did reveal evidence of generalized cerebellar atrophy. Initial genetic investigations including karyotype and Fragile X testing were normal.
At age 15 he presented back to genetics in the context of developing episodes consistent with hemiplegic migraine. As this time, given his constellation of features which include cerebellar ataxia, nystagmus, cranial nerve palsies, and a decline in IQ, a diagnosis of an atypical and severe early onset form of familial hemiplegic migraine was suspected. Genetic sequencing revealed the de novo missense change Y1384C in CACNA1A. In the 15 years since his genetic diagnosis, he has continued to follow a complex clinical course with cognitive decline, behavioural disturbances, disrupted sleep and yet no evidence of confirmed seizures. His care has been refractory to multiple medications.
One previous patient has also been reported in the literature with the same de novo variant. This individual was 33 years old at the time of publication [17] and has a similar history to the patient above with an uneventful birth history, cerebellar ataxia, nystagmus, global developmental delay, intellectual disability (estimated IQ of 40) and development of infrequent hemiplegic migraines beginning in childhood. These patients, and this specific variant, have been identified in reviews of FHM as being a variant of unique consequences, with an atypically severe phenotype [24].

The Y1384C mutation decreases Ca 2+ current density
To assess the functional effects of the missense mutation p.Y1824C, the mutation was introduced into the human Ca V 2.1 channel (splice isoforms + 47 and Δ47), and whole cell Ca 2+ currents from transiently transfected cells (Ca V 2.1 or Y1384C (+ 47 or Δ47) with Ca V β 4 and Ca V α 2 δ-1) were recorded using the patchclamp technique. Representative whole-cell Ca V 2.1 current recordings are shown in Fig. 2a. Figure 2b, c show the average current density-voltage relationships (peak current amplitude normalized by Cm) in response to 250 ms depolarizations from a holding potential (V h ) of − 100 mV. The current densities of the Y1384C mutant were consistently smaller across a range of voltages when compared to WT Ca V 2.1channel controls (Fig. 2b, c). In agreement with previous findings Ca V 2.1 (+ 47) channels exhibited larger peak current densities than Ca V 2.1 (Δ47) ( [10]; Fig. 2d). Y1384C mutants exhibited dramatically reduced peak current density indicating a marked loss of function (Fig. 2d). These data indicate that Y1384C mutants are either functionally inhibited or that they exhibit reduced trafficking to the cell surface. To discriminate between these two possibilities, we performed cell surface biotinylation experiments with Ca V 2.1 (Δ47) and Y1384C (Δ47) channels. As shown in Fig. 2e, the mutation did not affect the membrane expression of the channel, indicating that the decrease in current density is due to a biophysical effect. To determine if the Y1384C mutation influences inactivation kinetics, we measured the percentage of current inactivated at the end of a 250 ms test pulse. The mutation did not affect this parameter (Fig. 2e).

The Y1384C mutation modifies channel gating properties of Ca V 2.1 splice isoforms
To further study if Y1384C mutation had an effect on the functional properties of Ca V 2.1 variants, we analyzed the voltage-dependence of activation (Fig. 3). Slope factors were significantly increased in both splice variants in the presence of the mutation (Ca . Y1384C (Δ47) exhibited a shift of ~ 6.6 mV to more hyperpolarized potentials on its mean half-activation potential compared with Ca V 2.1 (Δ47) (Fig. 3a), and a similar effect was seen with Y1384C (+ 47) (Fig. 3d and Table 1), indicating a gain of function, that is however, offset by the reduced current densities reported in Fig. 2. Interestingly, several CACNA1A mutations located in the Cav2.1 channel pore (S5, S6 and linker) also mediate a shift to more hyperpolarized voltages and enhanced channel open probabilities [25,26].
To study if Y1384C mutation influences channel availability, the voltage dependence of inactivation was evaluated using 5-s pre-pulses depolarizations from − 100 to 0 mV preceding a 140-ms test potential to − 5 mV. We observed a hyperpolarizing shift of the mean-half inactivation potential of 7.7 mV for Y1384C (Δ47; Fig. 3b and Table 1) and 10.9 mV for Y1384C (+ 47; Fig. 3e and Table 1) without any changes in the slope factor (Table 1). Collectively, these data indicate that the Y1384C mutation produces significant alterations of Ca V 2.1 channels gating, by altering both activation and inactivation of the Ca V 2.1 (Δ47) and Ca V 2.1 (+ 47) channels. Since both activation and inactivation curves were shifted in Y1384C channels, we analyzed the window current of Ca V 2.1 channels. We observed an increase in the window current generated by Y1384C channels (Fig. 3c, f ). The Y1384C (Δ47) variant showed a 41% increase in the area under the curves whereas Y1384C (+ 47) variant exhibited a 21% increase. In addition, both mutants exhibited a hyperpolarized shift of the peak-voltage of the window current compared to their Ca V 2.1 control channel (Y1384C (Δ47): 9 mV and Y1384C (+ 47): 7.5 mV). Altogether, these sets of data indicate that Y1384C mutants exhibit a higher persistent activity which can be translated to a greater Ca 2+ current over a physiologically relevant membrane potentials near the resting potential.

Structural modeling of the Y1384 variant
The Y1384C mutation is located in the transmembrane segment 5 of the third domain of the Ca V 2.1α 1 subunit. To understand how a change from tyrosine to cysteine can modify channel biophysical properties, we generated a homology model of Ca V 2.1α 1 (Fig. 5a). This model was used for structural analysis, revealing that the tyrosine substitution by cysteine leads to the expansion of cavity volume by 166.752 Å 3 , and this increase is consistent with structural damage. Figure 5b shows the modifications (black) of some amino acids positions when tyrosine (red) is mutated to cysteine (blue). Finally, we wanted to investigate if the channels have electrostatic potential changes because of the presence of the Y1384C mutation.  Figure 5c, shows apparent changes in the extracellular link (black arrows) which were probably allosterically induced by the structural damage caused by the amino acid substitution.

Discussion
CACNA1A pathogenic variants have been classically associated with several disorders, including episodic ataxia type 2, spinocerebellar ataxia type 6 and hemiplegic migraine (familiar and sporadic [14,27]). Only few cases of early onset ataxia, permanent ataxia, or earlyonset cerebellar symptoms consistent with congenital ataxia have been associated with (de novo) CACNA1A pathogenic variants [13-15, 18, 28-32]. Our patient and the patient identified in Vahedi et al. [17] were two of the earliest reported examples of de novo developmental type disorders that are now increasingly recognized (see [33,34]). Congenital ataxias (CA) represent 10% of nonprogressive infantile encephalopathies [35]. Patients with CA present neonatal hypotonia and motor delay, and during the first years of life progressive ataxia [35].
Here, we studied the CACNA1A p.Y1384C mutation found in two individuals with congenital ataxia, early onset cerebellar atrophy, sporadic hemiplegic migraine and intellectual disability [17,18]. The symptoms in the patient reported here have been particularly challenging to manage. Given the emerging knowledge of the role in CACNA1A in various developmental encephalopathies, and the vision of provided precision care to rare disease patients, it is important to further understand the pathological consequences of individual rare variants. Based on our modeling work, the Y1384C mutation is predicted to compromise channel function. Cav3 channels contain a cysteine in the position corresponding to tyrosine 1384 in Cav2 channels. This may perhaps explain why the Y1384C mutation leads to a hyperpolarizing shift in half activation voltage, leading to a gain of function. Based on our modeling, it is unlikely that this involves the formation of disulfide bridges. On the other hand, it is possible electrostatic potential changes induced by the mutation may affect permeation properties, leading to reduced currents. Remarkably, in a study where two different FMH-1 CACNA1A mutations were modeled (W1684R and V1696I which are both located in the S4 domain), the former caused electrostatic potential changes whereas the latter did not [36]. Patients bearing the W1684R mutation, but not the V1696I mutation, have been shown to exhibit cerebellar ataxia as part of their phenotype [37]. Y1384C channels present a loss of function in current density and a gain of function in gating properties, similar to other mutations located at the inner pore [14]. Of those two effects, the loss of function appears to dominate, as clearly seen from the currentdensity voltage relations in Fig. 2. Five other mutations associated with congenital ataxia located at the pore domain have been examined in functional studies (T666M, I1811L, D715E, ΔF1502 and V1396M [26,[38][39][40][41]). T666M (located at the selectively filter in domain III) is one of the most reported variants. This mutation induces a gain of function shift in the activation curve. Patients present normal to mild intellectual disabilities, mild or moderate ataxia and episodic coma. Normal posterior fossa structures to cerebellar atrophy restricted to the vermis or generalized have been reported [29,42]. On the other hand, ΔF1502 (located at the S6 segment domain III) presents a reduction in current density, however when channels are stimulated with single or trains of action potentials, ΔF1502 exhibits an increase in Ca 2+ influx after stimulation. Patients with this mutation have chronic cerebellar syndrome with acute HM events and epileptic seizures [40]. Although gain of function in gating parameters are consistent among these mutants, it is important to take into account that associated pathologies are a reflection of the balance of the expression of the mutations between excitatory and inhibitory circuits. Indeed, alternative splicing [11,12], interaction with different ancillary subunits [43] and other regulatory and structural proteins [11,44] can generate different pools of channels located in different neurons or even in the same synapse.
It is important to note that CACNA1A mutations can give rise to a wide spectrum of FHM1 severity. For example, the well documented R192Q variant is relatively mild whereas the S218L variant is severe and can lead to fatal edema [45], and in homozygous S218L mice, sudden unexpected death in epilepsy (SUDEP) is observed [46]. Interestingly, like with the Y1384C mutant, both variants have similar gain of function effects on half activation voltage [12]. They however differ in their kinetics of recovery from inactivation, with the S218L variant exhibiting more rapid recovery, similar to what we observed here. The amount of whole cell current remaining at the end of repetitive trains of action potentials was significantly depressed in the S218L variant compared to WT or R192Q channels [12], consistent with a loss of function. These types of loss of function mixed with gain of function characteristics complicate treatment strategies, and make it difficult to correlate severity of symptoms with observed biophysical changes. Such changes can potentially disrupt Ca 2+ homeostasis in the cerebellum that can lead to congenital ataxia. For example, mutations on plasma membrane Ca 2+ ATPases, abundantly expressed in Purkinje cells and granule cells, show a reduced capacity to extrude Ca 2+ , disrupting basal Ca 2+ levels and/or Ca 2+ signalling, resulting in altered synaptic efficiency and promoting hyperexcitability that gives rise to an ataxic phenotype [47][48][49][50][51]. Although there is no general mechanism established by which alterations of Ca 2+ homeostasis in the cerebellum causes congenital ataxia, characterizing CACNA1A-linked mutations and their consequences in the cerebellar network may be important considerations for therapeutic interventions.