Cation leak underlies neuronal excitability in an HCN1 developmental and epileptic encephalopathy

Abstract

Pathogenic variants in HCN1 are associated with developmental and epileptic encephalopathies. The recurrent de novo HCN1 M305L pathogenic variant is associated with severe developmental impairment and drug-resistant epilepsy. We engineered the homologue Hcn1 M294L heterozygous knock-in (Hcn1M294L) mouse to explore the disease mechanism underlying an HCN1 developmental and epileptic encephalopathy. The Hcn1M294L mouse recapitulated the phenotypic features of patients with the HCN1 M305L variant, including spontaneous seizures and a learning deficit. Active epileptiform spiking on the electrocorticogram and morphological markers typical of rodent seizure models were observed in the Hcn1M294L mouse. Lamotrigine exacerbated seizures and increased spiking, whereas sodium valproate reduced spiking, mirroring drug responses reported in a patient with this variant. Functional analysis in Xenopus laevis oocytes and layer V somatosensory cortical pyramidal neurons in ex vivo tissue revealed a loss of voltage dependence for the disease variant resulting in a constitutively open channel that allowed for cation 'leak' at depolarized membrane potentials. Consequently, Hcn1M294L layer V somatosensory cortical pyramidal neurons were significantly depolarized at rest. These neurons adapted through a depolarizing shift in action potential threshold. Despite this compensation, layer V somatosensory cortical pyramidal neurons fired action potentials more readily from rest. A similar depolarized resting potential and left-shift in rheobase was observed for CA1 hippocampal pyramidal neurons. The Hcn1M294L mouse provides insight into the pathological mechanisms underlying hyperexcitability in HCN1 developmental and epileptic encephalopathy, as well as being a preclinical model with strong construct and face validity, on which potential treatments can be tested.

Keywords: HCN channels; developmental and epileptic encephalopathy; epilepsy; genetic mouse model; ion channel.

Figure 1

Size and survival of Hcn1^M294L mice. (A) Hcn1^M294L mice (left) were significantly smaller than their wild-type (WT) littermates (right). (B) Body weight of Hcn1^M294L mice and wild-type littermates over three time points in the 3 weeks post-weaning (n = 17 wild-type, 16 Hcn1^M294L). *P < 0.001 for all three time points. (C) Nissl staining of whole brain slice from wild-type (left) and Hcn1^M294L (right) mouse (scale bar = 520 µm). (D) There was a significant difference in survival of wild-type and Hcn1^M294L mice between weaning and P30 (n = 221 wild-type, 192 Hcn1^M294L), *P = 0.04 (Mantel-Cox log-rank test).

Figure 2

Hcn1^M294L mice displayed an epileptic phenotype. (A) Sample ECoG traces from a wild-type (WT) (top) and an Hcn1^M294L (bottom) mouse, showing spiking (indicated with an asterisk) in the Hcn1^M294L trace (scale bar = 3 s, 90 µV). Inset: Expanded view of an Hcn1^M294L spike (asterisk) showing epileptiform morphology (scale bar = 500 ms, 90 µV). (B) Similar spikes were not seen in ECoG traces from wild-type mice (n = 4 wild-type, 7 Hcn1^M294L), *P = 0.01. (C) Representative ECoG traces from an Hcn1^M294L mouse before (top) and after (bottom) administration of sodium valproate (200 mg/kg) (scale bar = 3 s, 90 µV). (D) Sodium valproate (200 mg/kg) caused a significant decrease in ECoG spike frequency in Hcn1^M294L mice (n = 11), *P = 0.005. (E) Representative ECoG traces from an Hcn1^M294L mouse before (top) and after (bottom) administration of lamotrigine (20 mg/kg) (scale bar = 3 s, 90 µV). (F) Lamotrigine (20 mg/kg) caused a significant increase in ECoG spike frequency in Hcn1^M294L mice (n = 11), *P = 0.002. (G) GFAP staining of primary somatosensory cortex of a wild-type (top) and an Hcn1^M294L (bottom) mouse, showing gliosis in the Hcn1^M294L animal (scale bar = 300 µm). (H) Nissl staining showing morphological changes (top row, arrow), upregulation of the major epilepsy marker NPY (middle row, arrow), and upregulation of the reactive gliosis marker GFAP (bottom row) in the hippocampus of Hcn1^M294L mice (right column) compared to wild-type (left column) (scale bar = 250 µm) (n = 6 wild-type, 6 Hcn1^M294L). Hippocampal gliosis and upregulation of NPY were seen in 6/6 Hcn1^M294L mice and 0/6 wild-type mice. (I) Kaplan-Meier curve showing Hcn1^M294L mice had a significantly shorter time to terminal seizure in the thermogenic seizure assay compared to wild-type littermates (n = 7 wild-type, 7 Hcn1^M294L), *P = 0.004. (J) Kaplan-Meier curve showing Hcn1^M294L mice had a significantly shorter time to tonic-clonic seizure in the subcutaneous PTZ assay compared to wild-type littermates (n = 9 wild-type, 10 Hcn1^M294L), *P < 0.0001.

Figure 3

Hcn1^M294L mice displayed significant behavioural differences from wild-type littermates. (A) Hcn1^M294L mice travelled significantly further than wild-type (WT) littermates in the open field locomotor assay over a 60-min period (n = 10 wild-type, 10 Hcn1^M294L), *P < 0.0001. (B) There were no significant differences between genotypes in the percentage of time spent in the novel arm of the Y maze (n = 10 wild-type, 10 Hcn1^M294L), P = 0.6. (C) Hcn1^M294L mice showed a significantly increased average latency to target in the Barnes maze assay (n = 10 wild-type, 10 Hcn1^M294L), *P = 0.01 on Day 7. (D) Hcn1^M294L mice spent significantly longer in the dark zone of the light-dark box (n = 10 wild-type, 10 Hcn1^M294L), *P = 0.002. (E) Hcn1^M294L mice spent significantly longer in the open arms of the elevated plus maze (n = 10 wild-type, 10 Hcn1^M294L), *P = 0.002. (F) Time spent interacting with an age- and sex-matched stranger mouse in the three-chamber social interaction assay. Female, but not male, Hcn1^M294L mice showed a deficit in social interaction (n = 5 wild-type males, 6 wild-type females, 4 Hcn1^M294L males, 5 Hcn1^M294L females), *P = 0.01.

Figure 4

Functional analysis of HCN1 M305L revealed significantly altered activation kinetics but unchanged cation selectivity. (A) Structure of human HCN1 showing location of M305 (purple spheres). The S4 helices of the voltage sensing domains are shown in blue and S5 helices shown in orange. Based on PDB 5u60 for human HCN1 in the depolarized (closed) conformation rendered using PyMOL (The PyMOL Molecular Graphics System, Version 2.3.4 Schrödinger, LLC.). (B) Representative voltage clamp data from oocytes expressing HCN1 wild-type (WT) (left); M305L (middle) and co-expressed WT+M305L (right) (scale bar = 1 s, 10 µA). Each dataset shows current traces in response to a series of voltage steps (inset) from the holding potential (−30 mV) to test potentials in the range −120 mV to +20 mV. Grey shaded boxed areas have been enlarged to show tail currents visible for wild-type and WT+M305L, but absent for M305L (scale bar = 0.1 s, 2 µA). Red trace corresponds to voltage step to −120 mV. Arrow highlights tail current present for depolarizing steps with WT+M305L. (C) Average of raw steady-state current for HCN1 WT (n = 14) and WT+M305L (n = 17) measured at end of test pulse to −120 mV. (D) Grouped current-voltage (I–V) data normalized to −100 mV shows markedly weaker inward rectification for WT+M305L (n = 4) co-expressing oocytes compared to WT (n = 5). Error bars are smaller than symbols. (E) Mean activation time constant τ₂^act obtained by fitting the time-dependent component of activating current from wild-type (n = 9) and WT+M305L (n = 10) oocytes with a two-exponential function. (F) Normalized I–V data for WT (left, n = 6) and M305L (right, n = 5) for different external ion substitutions. Data were normalized to the steady-state current in 100 K at −100 mV. Li⁺ or choline resulted in much smaller currents, no rectification, and resembled endogenous currents seen in non-injected oocytes (data not shown). (G) I–V data from F normalized to the steady-state current in 100 K at −100 mV, after subtraction of current in 100 choline. Black symbols = wild-type (n = 10); red symbols = M305L (n = 9). Filled and open symbols are with 100 K superfusion and 100 Na superfusion, respectively. Grey shading represents critical range of potentials for AP initiation and enlarged in inset for 100 Na superfusion.

Figure 5

I_h recorded from Hcn1^M294L layer V pyramidal neurons lacks voltage dependence. (A) Representative voltage clamp data from layer V neurons from wild-type (WT) (blue) and Hcn1^M294L (red) mice, at baseline (left, ‘pre-ZD’), following the application of ZD 7288 (centre, ‘post-ZD’), and I_h isolated by subtracting ZD 7288-sensitive current traces from baseline traces (right, ‘subtraction’) (scale bar = 1 s, 200 pA). Each dataset shows current traces in response to a series of voltage steps (inset) from the holding potential (−50 mV) to test potentials in the range −110 mV to −45 mV. (B) Current-voltage (I–V) data shows a depolarizing shift in the reversal potential of Hcn1^M294L neurons (n = 14) compared to wild-type (n = 12), *P < 0.0001. (C) I–V relationship for I_h from Hcn1^M294L neurons (n = 5 wild-type, 5 Hcn1^M294L) shows minimal rectification. (D) The activation time constant (tau) of I_h in Hcn1^M294L neurons is substantially faster and shows substantially weaker voltage dependence compared to wild-type (n = 5 wild-type, 5 Hcn1^M294L).

Figure 6

Layer V pyramidal neurons from Hcn1^M294L mice were hyperexcitable. (A) HCN1 protein staining of cortex from a wild-type mouse (top) and an Hcn1^M294L mouse (bottom), showing the expected predominant localization to the apical dendrites in both genotypes (scale bar = 100 µm). (B) Average of sag traces obtained from cells from wild-type mice (top, blue, n = 17) and cells from Hcn1^M294L mice (bottom, red, n = 22) (scale bar = 1 s, 5 mV). Layer V slice electrophysiology experiments were conducted on tissue from n = 7 wild-type and 8 Hcn1^M294L mice. (C) Sag is significantly reduced in layer V neurons from Hcn1^M294L mice (n = 17 wild-type, 22 Hcn1^M294L), *P = 0.03. (D) Example current clamp gap-free recordings from a layer V neuron from an Hcn1^M294L mouse (red) and from a wild-type mouse (blue) in the absence of any holding current. (E) Layer V neurons from Hcn1^M294L mice displayed a significantly depolarized resting membrane potential compared to layer V neurons from wild-type mice (n = 10 wild-type, 11 Hcn1^M294L), *P < 0.0001. (F) In the presence of the synaptic blocking drugs AP5, NBQX and picrotoxin, layer V neurons from Hcn1^M294L mice remained more depolarized than neurons from wild-type mice (n = 4 wild-type, 5 Hcn1^M294L), *P = 0.03. (G) The HCN channel blocking drug ZD 7288 caused a significant hyperpolarization of the RMP in both Hcn1^M294L neurons (n = 7, *P = 0.0008) and wild-type neurons (n = 7, *P = 0.0005). (H) Representative AP firing patterns in response to 2 s injections of +75 pA (top) and +200 pA (bottom) current in layer V neurons from wild-type (left, blue) and Hcn1^M294L (right, red) mice (scale bar = 1 s, 50 mV). (I) Summary of i–o data showed a left-shift in the input-output relationship in Hcn1^M294L layer V neurons compared to wild-type neurons (n = 10 wild-type, 11 Hcn1^M294L). (J) Rheobase was left-shifted in layer V neurons from Hcn1^M294L mice (n = 10 wild-type, 11 Hcn1^M294L), *P = 0.01. (K) There was no significant difference in gain, as measured by area under the curve in the 15 sweeps starting from the sweep immediately prior to rheobase, between layer V neurons from wild-type and Hcn1^M294L mice (n = 9 wild-type, 11 Hcn1^M294L), P = 0.7.

Figure 7

Input resistance and AP properties of layer V pyramidal neurons from Hcn1^M294L mice. (A) There was no significant difference in input resistance between wild-type (WT) and Hcn1^M294L neurons when measured from rest (n = 10 wild-type, 9 Hcn1^M294L), P = 0.5. (B) There was no significant difference in input resistance between wild-type and Hcn1^M294L neurons at a holding potential of −70 mV (n = 16 wild-type, 21 Hcn1^M294L), P = 0.8. (C) Hcn1^M294L neurons had a significantly lower input resistance than wild-type neurons at a holding potential of −50 mV (n = 5 wild-type, 5 Hcn1^M294L), *P = 0.02. (D) After the application of ZD 7288, there was no significant difference in input resistance between WT and Hcn1^M294L neurons at a holding potential of −50 mV (n = 4 wild-type, 4 Hcn1^M294L), P = 0.8. (E) Representative raw traces of the first AP fired at rheobase in a layer V neuron from a wild-type mouse (blue) and from an Hcn1^M294L mouse (red). (F) Layer V neurons from Hcn1^M294L mice had a significantly depolarized threshold for AP firing (n = 10 wild-type, 10 Hcn1^M294L), *P = 0.003. (G) Layer V neurons from Hcn1^M294L mice had a significantly reduced AP amplitude (n = 10 wild-type, 10 Hcn1^M294L), *P = 0.01. (H) Expression of Scn2a mRNA in cortical tissue was significantly reduced in Hcn1^M294L mice (n = 5 wild-type, 5 Hcn1^M294L), *P = 0.008. Expression of Scn8a mRNA in cortical tissue was significantly reduced in Hcn1^M294L mice (n = 5 wild-type, 5 Hcn1^M294L), *P = 0.04.

Figure 8

CA1 neurons from Hcn1^M294L mice were hyperexcitable. (A) Average of sag traces obtained from cells from wild-type (WT) mice (top, blue, n = 12) and cells from Hcn1^M294L mice (bottom, red, n = 15) (scale bar = 1 s, 10 mV). Slice electrophysiology experiments were conducted on tissue from n = 2 wild-type and 2 Hcn1^M294L mice. (B) Sag is significantly reduced in CA1 neurons from Hcn1^M294L mice (n = 12 wild-type, 15 Hcn1^M294L), *P < 0.0001. (C) Example current clamp gap-free recordings from a CA1 neuron from an Hcn1^M294L mouse (red) and from a wild-type mouse (blue) in the absence of any holding current. (D) CA1 neurons from Hcn1^M294L mice displayed a significantly depolarized resting membrane potential compared to CA1 neurons from wild-type mice (n = 14 wild-type, 15 Hcn1^M294L), *P < 0.0001. (E) Representative AP firing patterns in response to 2 s injections of +50 pA (top) and +250 pA (bottom) current in CA1 neurons from wild-type (left, blue) and Hcn1^M294L (right, red) mice (scale bar = 1 s, 50 mV). (F) Summary of i–o data showed a left-shift in rheobase and an earlier firing collapse in Hcn1^M294L CA1 neurons compared to wild-type neurons (n = 14 wild-type, 15 Hcn1^M294L). (G) Rheobase was left-shifted in CA1 neurons from Hcn1^M294L mice (n = 14 wild-type, 15 Hcn1^M294L), *P < 0.0001. (H) There was no significant difference in input resistance between wild-type and Hcn1^M294L CA1 neurons when measured from rest (n = 14 wild-type, 14 Hcn1^M294L), P = 0.6.