Cation leak: a common functional defect causing HCN1 developmental and epileptic encephalopathy

Abstract Pathogenic variants in HCN1 are an established cause of developmental and epileptic encephalopathy (DEE). To date, the stratification of patients with HCN1-DEE based on the biophysical consequence on channel function of a given variant has not been possible. Here, we analysed data from eleven patients carrying seven different de novo HCN1 pathogenic variants located in the transmembrane domains of the protein. All patients were diagnosed with severe disease including epilepsy and intellectual disability. The functional properties of the seven HCN1 pathogenic variants were assessed using two-electrode voltage-clamp recordings in Xenopus oocytes. All seven variants showed a significantly larger instantaneous current consistent with cation leak. The impact of each variant on other biophysical properties was variable, including changes in the half activation voltage and activation and deactivation kinetics. These data suggest that cation leak is an important pathogenic mechanism in HCN1-DEE. Furthermore, published mouse model and clinical case reports suggest that seizures are exacerbated by sodium channel blockers in patients with HCN1 variants that cause cation leak. Stratification of patients based on their ‘cation leak’ biophysical phenotype may therefore provide key information to guide clinical management of individuals with HCN1-DEE.


Introduction
The developmental and epileptic encephalopathies (DEEs) are the most severe subgroup of epilepsies. They are characterised by drug-resistant seizures and epileptiform activity that result in developmental slowing or regression. 1 Pathogenic variation in over eight hundred genes are now established causes of DEEs. 2 Natural history studies are critical to establish the phenotypic spectrum and evolution of a given genetic DEE, including severity and range of epilepsy phenotypes, time course and pharmaco-responsiveness, and associated multimorbidities. 3 Understanding the spectrum of natural histories of a genetic DEE can be further interpreted by considering the functional consequence of a pathogenic variant. This is perhaps best exemplified by DEEs caused by pathogenic variants in the sodium channel alpha-2 subunit gene, SCN2A, where stratification into variants causing either gain-of-function (GOF) or loss-of-function (LOF) has some ability to predict the natural history course including pharmacosensitivity. [4][5][6][7] The categorisation of SCN2A DEE patients also carries critical implications for precision medicine approaches. Patient assignment based on the biophysical impact has also been established for other ion channel genes, including those encoding other sodium channels and GABA A receptors. 8,9 There is thus a crucial need to understand the functional consequence of a pathogenic variant in the development of precision medicines.
Pathogenic variants in HCN1 are an established cause of DEE. [10][11][12] Variants occurring in the transmembrane domains of HCN1 channels are more likely to cause severe disease. 10 However, no clear correlation between the impact of a variant on the biophysical properties of the HCN1 protein and disease outcome has been established. 13 This may be partially due to the incorrect assignment of GOF or LOF to a given variant. For example, the original classification of the recurrent HCN1 M305L variant was LOF. 10 We have since established cation leak as the pathophysiological basis of HCN1-DEE disease for the HCN1 M305L variant. 14,15 Our functional analysis in Xenopus oocytes revealed an uncoupling between the pore and voltage sensor domains, rendering the channel in a mostly open state. 14,15 The cation leak phenotype was confirmed in mouse L5 pyramidal neurons carrying the homolog of the HCN1 M305L variant. 14 Despite this, the HCN1 M305L variant has continued to be considered LOF by other research groups. 11 Two technical issues could explain this. Firstly, 'leaky' cells are historically discarded in electrophysiological experiments as they are deemed to be of poor quality. Secondly, leak subtraction is frequently applied to electrophysiology data to remove background 'leak' predominantly through potassium channels. This has important ramifications for measuring changes in channel function in which constitutively open channels contribute significantly to the biophysical impact of a pathogenic variant. Here, we report functional analysis of two unstudied HCN1 variants, together with reanalysis of five variants with some published functional data. [10][11][12][16][17][18] All pathogenic variants were in the transmembrane domains of the HCN1 protein. A cation leak was established for all, suggesting that there is a common pathogenic mechanism underlying a subset of HCN1 epilepsy.

Selection of variants
We studied the biophysical properties of seven different de novo HCN1 pathogenic variants located in the transmembrane domains of the HCN1 protein that are carried by eleven patients.

Patient consent
The parents of new patient (

Site-directed mutagenesis and in vitro cRNA preparation
cDNA encoding a full-length transcript of wild-type human HCN1 (RefSeq NM_021072.4 Ensemble database) was subcloned into the pGEMHE-MCS vector. Site-directed mutagenesis to create human HCN1 variants was completed by GenScript Biotech (Piscataway, NJ, USA). All clones were verified by Sanger sequencing then linearized with NheI-HF (New England Biolabs) and purified using QIAquick PCR Purification Kit (QIAGEN). In vitro synthesis of cRNA was performed using linearized cDNA template and the mMessage mMachine® T7 transcription kit (Ambion, Thermo Fisher Scientific, Waltham, MA) and purified using RNeasy Mini Kit (QIAGEN). RNA integrity was assessed using NanoDrop Spectrophotometer and gel electrophoresis. cRNAs were stored at -80°C.

Xenopus oocyte electrophysiology Oocyte extraction
Adult female Xenopus laevis frogs were anaesthetized with 1.3 mg/ml tricaine methanosulfonate and oocytes were surgically removed via a small incision in the abdomen. Oocytes were defolliculated with 1.5 mg/ml collagenase for 2 h and rinsed with OR-2 solution (in mmol/L: 82.5 NaCl, 2 KCl, 1 MgCl 2 .6H 2 O, 5 HEPES, pH 7.4). Healthy mature oocytes stage V and VI were isolated for experiments.

Channel expression
cRNAs of 100 ng/µL coding for HCN1 were manually injected into the oocytes to give a total injection volume of 50 nL. Injected oocytes were maintained in ND96 storage solution (in mmol/L: 96 NaCl, 2 KCl, 1 MgCl 2 .6H 2 0, 1.8 CaCl 2 .2H 2 O, 5 HEPES, 50 mg/L gentamicin, pH 7.4) at 17°C for two to three days to allow translation and trafficking of channels before experimentation.

Two-electrode voltage-clamp electrophysiology
Standard two-electrode voltage-clamp hardware was used (TEC-05X or TEC10X, NPI, Tamm, Germany) with series resistance (R s ) compensation applied to ensure clamp accuracy when oocytes exhibited high functional expression. Oocytes were impaled with microelectrodes containing 3 mol/L KCl and with an input resistance between 0.2 and 1.5 MΩ. During experiments oocytes were continually perfused with high K + solution (in mmol/L: 100 KCl, 1.8 CaCl 2 , 1 MgCl 2 , 10 HEPES, pH 7.4). All recordings were performed at 18-20°C. Voltage clamp control and data acquisition were obtained using pCLAMP8.10 software (Molecular Devices, USA). Oocytes were clamped at -30 mV holding potential and current-voltage relationships were generated using a voltage step protocol with incremental 10 mV steps from -120 mV to +20 mV for 2.5 s. Data were sampled at 200 µs/point and low pass filtered at 500 Hz. CsCl was dissolved in high K + solution at a concentration of 10 mM before being applied during experiments. Oocytes used for experiments were selected if stable and had expression of the relevant channel.

Electrophysiological data analysis
Data were analysed using Prism 8.4.2 (GraphPad, USA) and Clampfit 10.4 (Molecular Devices, USA). The total steadystate current was measured over an approximately 200 ms interval at the end of the test pulse. For tail current analysis, baseline correction was applied after the current had reached a steady-state and the instantaneous current was estimated at the start of the repoad settled. This current was measured and fit with a form of the Boltzmann equation: where I max tail is the maximum instantaneous tail current and I off tail is the offset reported by the fit (typically =0 with baseline correction), V 0.5 is the mid-point voltage, z the apparent valence of the charge moved, and kT/e = 25.3 mV at 20°C. To characterise the channel activation kinetics, we fit a single exponential function to the activating current commencing after the initial inflexion. A double exponential fit (see Hung et al. 2021) was found to report reliable fits only for the WT and some of the constructs at extreme hyperpolarizing potentials. To characterise the channel deactivation kinetics, a single exponential fit was found to adequately describe the WT and WT+ variant channels for the monotonic relaxation phase.

Statistical analysis
Data were analysed with Clampfit10.4 (Molecular Devices, USA) and Prism 8.4.2 (GraphPad, USA). Standard one-way ANOVA with Dunnett's post-hoc correction was used for statistical comparisons to wild-type values. Significance was set at P < 0.05. Data were collected from at least two different batches of oocytes. All data points are shown as mean ± SEM unless otherwise stated.

Clinical phenotypes of patients with pathogenic HCN1 variants in the transmembrane domains
We present the phenotypic data from eleven patients harboring seven different de novo HCN1 pathogenic variants located in the transmembrane domains of the protein (Fig. 1A, Table 1). We report one new patient (patient 8) carrying the recurrent HCN1 A387S variant and ten patients who have been previously published. [10][11][12][16][17][18] Table 1 summarizes the main clinical features of these patients. HCN1-DEE was severe in all individuals, with six patients having early infantile DEE (which now includes neonatal onset DEE) and five having DEE. 19 Seizure onset occurred at a median age of 2.5 months (range 30 h to 8 months). At onset, seizures were triggered by fever in 7/11 patients. All patients developed additional seizure types, including focal, generalized and focal to bilateral tonic-clonic, myoclonic and absence seizures. EEG analysis typically showed multifocal epileptiform activity and diffuse slowing. All patients had intellectual disability, which was typically severe. Autistic features and behavioural problems were reported in some individuals. Two patients had microcephaly. Three patients died at ages between 12 and 15 months.

Leak current is a common biophysical feature of all transmembrane de novo pathogenic HCN1 variants tested
Two-electrode voltage-clamp recordings were made from Xenopus laevis oocytes to investigate the functional consequences of seven transmembrane-domain HCN1 variants. These included five published variants (S272P, M305L, I380F, G391D, and S399P) and two published variants that have not been previously functionally characterized (A387S and I380N) (Fig. 1A). A 50:50 mix of each HCN1 pathogenic variant RNA with WT HCN1 RNA was injected into oocytes to mimic their heterozygous status in patients. All five previously characterized variants displayed recordable currents (Fig. 1B, Supplementary Fig. 1). Cs + is a known extracellular inorganic blocker of HCN channels and was used to distinguish between leak from endogenous channels and HCN channels. 20,21 Cs + (10 mM) effectively blocked WT and all WT+ variant HCN1-mediated currents (Fig. 1B, Supplementary Fig. 1). All subsequent analysis was completed on Cs + -subtracted currents.
An instantaneous jump in the voltage step-induced current is a measure of channels that are already open at the -30 mV holding potential. WT HCN1 channels generate a robust voltage-dependent current that can be measured at steady-state (I ss ) with minimal instantaneous current (I inst ), suggesting that the majority of channels are closed at a holding potential of -30 mV (Fig. 1C, arrow). In contrast, the S272P, M305L, I380F, G391D, and S399P HCN1 pathogenic variants show marked I inst reflective of mutant channels being open at the -30 mV holding potential (Fig. 1C,  arrows). Similarly, co-expression of the newly functionally characterized A387S and I380N pathogenic variants with WT HCN1 also show marked I inst (Fig. 1D, arrows). The average normalized I inst /I ss highlights that I inst is a significant proportion of the elicited current for all transmembrane HCN1 pathogenic variants tested (Fig. 1E). This highlights that all transmembrane HCN1 variants studied had a significant cation leak.

Impact of transmembrane de novo pathogenic HCN1 variants on other biophysical parameters
Further analysis reveals that HCN1 pathogenic variants impact on other biophysical parameters. The current-voltage relationships generated from normalised I ss are presented in Fig. 2A. The shaded box around -50 mV highlights larger currents for all variants, consistent with the cation leak described above. The impact of HCN1 pathogenic variant mediated I ss at -120 mV normalised to within batch WT I ss reveals that the majority of variants reduced expression levels in oocytes, although this was somewhat variable and may relate to variability inherent to the oocyte expression system (Fig. 2B). Current-voltage relationships generated from tail currents provide a measure of the channel open probability as a function of membrane potential (Fig. 2C). Fitting these data with a Boltzmann function (see Materials and Methods) revealed that the voltage of half activation (V 0.5 ) was mainly right shifted (Fig. 2D). This resulted in an increase in the probability of opening at -50 mV (P open ) for the majority of WT+ variant studied (Fig. 2E). The slope (E) Instantaneous current (normalised to steady-state current) at -100 mV for HCN1 WT and co-expressed WT + S272P, WT + M305L, WT + I380N, WT + I380F, WT + A387S, WT + G391D, and WT + S399P. P < 0.05 were considered significant and denoted *. Data were compared using one-way ANOVA with Dunnett's post-hoc, compared to WT (See Supplementary Table 1 for detailed analysis and exact P-values). of the current-voltage relationship (z) was generally less steep (Fig. 2F). The impact of pathogenic HCN1 variants on activation and deactivation was also variable with the time course mostly faster than HCN1 WT channels ( Fig. 2G and H). In summary, all transmembrane domain pathogenic HCN1 variants studied caused changes in one or more biophysical characteristics that increased the probability that the mutant channel would be open at membrane potentials where HCN1 WT channels would normally be closed, resulting in a cation leak.

Discussion
Pathogenic variants in HCN1 have emerged as an important cause of DEE. [10][11][12] Marini et al. reported that HCN1 variants associated with severe phenotypes tend to cluster within or close to the transmembrane domains. However, there has not been a clear functional correlate of the severe phenotype of HCN1-DEE. 13 Here we report that seven de novo HCN1 pathogenic variants which result in changes to amino acids within the transmembrane domains share a common mechanism of cation leak. Each of the eleven patients harbouring cation leak HCN1 variants have DEEs with severe developmental impairment and drug-resistant seizures. [10][11][12][16][17][18] We propose that HCN1 pathogenic variants that result in cation leak are an important cause of HCN1-DEE. Knock-in mouse models of HCN1-DEE based on human pathogenic variants have provided insight into the cellular and network basis of disease. 14,22 The Hcn1 M294L mouse model (homolog of HCN1 M305L) recapitulates the disease phenotypes seen in patients, including spontaneous seizures and cognitive deficits. 14 Whole-cell voltage clamp recordings from L5 pyramidal neurons reveal an HCN channelmediated cation leak similar to that reported in the oocyte expression system. 14,15 This leads to a more depolarised resting membrane potential in both L5 and CA1 pyramidal neurons, taking them closer to firing threshold. 14 A similar depolarised mechanism is reported for CA1 pyramidal neurons in the Hcn1 G380D mouse model (homolog of HCN1 G391D). 22 Importantly, both the Hcn1 M294L and Hcn1 G380D HCN1-DEE mouse models respond to standard anti-seizure medications (ASMs) in a similar manner. 14,22,23 This includes seizure exacerbation induced by lamotrigine and phenytoin, both sodium channel blocking ASMs. 14,22 In contrast, valproate reduced excitability on EEG (reduced spiking) in the Hcn1 M294L mouse and did not exacerbate seizures in the Hcn1 G380D mouse model. 14,22 This mirrors reports from the parents of children with HCN1-DEE who carry the pathogenic variants modeled in the respective mice including seizure exacerbation with lamotrigine and phenytoin and some efficacy with valproate. 14,22 Furthermore, sodium channel blocking ASMs also exacerbated seizures in the Hcn1 M142I mouse (homolog of HCN1 M153I). 22 Porro et al. reported a significant right shift in the voltage of half-activation of the M153I variant, increasing the probability of channels being open at depolarised potentials which would contribute to a cation leak. 13 Collectively, these data suggest that sodium channel blocking ASMs should be avoided in HCN1-DEE where pathogenic variants cause cation leak. These data also have implications for precision medicine approaches with strategies aimed at reducing cation leak likely to be beneficial for patients with HCN1-DEE.
HCN1 pathogenic variants are associated with a broad phenotypic spectrum including milder phenotypes such as idiopathic generalised epilepsies and genetic epilepsy with febrile seizures plus. 10 Our current functional data focuses on HCN1 pathogenic variants that cause cation leak associated with HCN1-DEE. Our current knowledge of HCN1 variants that do not cause cation leak is limited. Future studies investigating biophysical properties of HCN1 pathogenic variants associated with milder phenotypes are needed, included rodent and stem cell models, complemented by pharmacological studies.
Interestingly, there is phenotypic variability in patients with HCN1-DEE which correlates in part with the severity of the cation leak. The basis of these differences is likely to be complex. Factors could include subtle differences in other biophysical properties and variant-dependent differences in channel trafficking, as well as patients' genetic background.
The biophysical mechanisms underlying cation leak caused by HCN1 pathogenic variants are likely to be multifactorial. A large depolarising shift in the activation voltage as described above for the HCN1 M153I variant will result in more channels being open at depolarised potentials and contribute to an inward cation current. 10,13 Additionally, our work on the HCN1 M305L variant suggests a decoupling of the voltage-sensor and pore domain resulting in a constitutively open pore. 14,15 Recent biophysical models of voltage to pore coupling in HCN1 suggest that the pore is essentially 'spring-loaded', and that the movement of the voltage sensor frees space for the channel to open. 24,25 We, therefore, predict that the HCN1 pathogenic variants destabilise a 'scaffold', which would normally keep the channel closed at depolarized potentials, resulting in the uncoupling of voltage to pore domains and leading to a cation leak. Consistent with this, in addition to M305L, the variants I380N/F, G391D, and A387S also show reduced inward rectification at depolarised potentials. Although we have focused on transmembrane HCN1 variants, variants in other regions of the protein are also likely to cause changes that result in increased cation flow at more depolarised potentials. Additionally, future work is needed to see how HCN1 variants impact heteromeric channels including those formed with HCN2 channels and whether these data can further refine the genotype-phenotype relationship in HCN1-DEE.
For the DEEs, understanding the impact of each patient's pathogenic variant is critical. This not only includes the gene itself, but also how specific pathogenic variants impact protein function. The correct assignment of patients into functional groups has distinct ramifications for diagnosis and optimisation of therapy. This includes use of current ASMs and the development of precision medicines. Here, we identify cation leak as an important cause of HCN1-DEE. Precision therapies targeting cation leak may provide a common therapeutic mechanism for patients with this severe DEE.

Supplementary material
Supplementary material is available at Brain Communications online.