Neddylation stabilizes Na_v1.1 to maintain interneuron excitability and prevent seizures in murine epilepsy models

In the CNS, proper neural activities between excitatory pyramidal neurons (PyNs) and inhibitory interneurons (INs) are critical to brain function. In particular, parvalbumin-positive (PV⁺) interneurons (PVINs) are a major source of inhibitory signal in maintaining excitation-inhibition (E-I) balance. PVINs fire at high frequency and, via feedback and feedforward inhibitions, are necessary for the generation of local circuit oscillations and connectivities or synchronization between brain regions (1–6).The excitability of PVINs requires Na_v1.1, the α subunit of voltage-gated sodium channel that is specifically expressed in PVINs (7–13). Na_v1.1 deficiency reduces PVIN excitability, increases the threshold of action potential (AP), and decreases AP amplitudes (7, 8, 12, 14–16). On the other hand, PVINs are controlled by a heterogeneous group of voltage-gated potassium channels (such as K_v3.1, K_v3.2, K_v3.3, and KCNQ2/3) whose mutations also change their excitability (17–19). In accordance with this, mutations in Na_v1.1 channels have been identified in patients with mild as well as severe forms of epilepsy, including severe myoclonic epilepsy of infancy (SMEI or Dravet syndrome) as well as those with generalized epilepsy with febrile seizures plus (GEFS⁺) (20–22). Studies of mouse models with Na_v1.1 mutations or deletions suggest that epilepsy may be caused by impaired firing of GABAergic INs (7, 12, 16, 23–25). A unified loss-of-function hypothesis posits that mild impairment of Na_v1.1 functions causes febrile seizures, intermediate loss-of-function gives rise to GEFS⁺ epilepsy, and severe impairment causes the intractable seizures and comorbidities of SMEI (26). However, in contrast with the impact of Na_v1.1 mutations on channel properties, less is known about mechanisms in regulating Na_v1.1’s stability or expression. Nevertheless, mutations could be pathogenic via reducing surface expression caused by aberrant folding, trafficking, or degradation (27, 28).

Neddylation is a chemical reaction in which the ubiquitin-like protein neural precursor cell expressed, developmentally downregulated 8 (Nedd8) is conjugated to the lysine of substrate proteins by its C-terminal glycine (29). Like ubiquitination, neddylation requires Nedd8-specific E1, a heterodimer of NAE1 (also known as APP-BP1) and UBA3, which together activate NEDD8 in an ATP-dependent manner, UBC12 (a Nedd8 E2 enzyme), and substrate-specific E3 ligases. Neddylated proteins could be deneddylated by NEDD8-specific proteases, such as the COP9 signalosome (CSN) and NEDP1 (also known as SENP8; ref. 30). Nedd8 is evolutionarily highly conserved and ubiquitously expressed in many cell types (31, 32). Neddylation regulates a variety of cellular processes, including gene transcription, cell division and differentiation, ribosome biogenesis, apoptosis, and proteolysis (33–35). However, the function of neddylation in the brain is less clear, although it has been recently implicated in the development of the neuromuscular junction and excitatory synapses onto PyNs (36, 37).

In this study, we determined whether the development and function of INs is regulated by neddylation. To this end, we mutated NAE1, an obligatory subunit of the only identified Nedd8 E1 enzyme, specifically in PVINs. Unexpectedly, mutant (mt) mice displayed spontaneous epileptic seizures, impaired inhibitory synaptic transmission, and decreased excitability. Yet morphological studies suggest that the mutation had no apparent effect on the number of PVINs in the cortex and hippocampus. Remarkably, the release of GABA, but not glutamate, was reduced in mt mice, which was associated with reduced excitability of PVINs, suggesting that neddylation is critical to controlling the excitability of INs. Further molecular mechanistic studies identified Na_v1.1 as a target of Nae1 mutation, which becomes unstable in the absence of neddylation. This notion was supported by proteomic analysis, which also reveals abnormality in synapses and metabolic pathways. Finally, elevated PVIN excitability was attenuated by rescue expression of Na_v1.1 in Nae1 mt mice. Together, the results demonstrate a role of neddylation for Na_v1.1 stability and GABAergic transmission.

Ataxia and spontaneous epileptic seizures in PV-specific Nae1-deficient mice. To investigate the potential role of neddylation in PVINs, we crossed floxed Nae1 mice (38) with PV-Cre mice in which Cre expression was under the control of the PV promoter (ref. 39 and Figure 1A). NAE1 expression in the brain of PV-Nae1^–/– mice was characterized. Quantitative real-time PCR (RT-PCR) and Western blot (WB) analyses of hippocampal tissues showed a trend of reduction of NAE1 mRNA and protein between PV-Nae1^–/– and PV-Cre mice, but the reduction was statistically insignificant (Supplemental Figure 1, A–C; supplemental material available online with this article; https://doi.org/10.1172/JCI136956DS1). This was not unexpected because NAE1 is expressed in many types of cells in the brain, including PyN and astrocytes. In accordance with this, immunostaining indicated that NAE1 was reduced in PVINs of hippocampal sections, while no change was observed in PV-negative cells (Figure 1B), suggesting that NAE1 was specifically reduced in PVINs. To test this notion further, cortical neurons were cultured from PV-Cre::Ai9 mice (referred to as PV-Ai9), where tdTomato was specifically expressed in PVINs (40). NAE1 was detectable in PV-positive cells from PV-Ai9 mice, but not those from PV-Nae1^–/–-Ai9 mice. NAE1 was also detectable in PV-negative cells, but at similar levels between the 2 genotypes (Supplemental Figure 1, D and E). Together, these results demonstrate that NAE1 was specifically ablated in PVINs in PV-Nae1^–/– mice. PV-Nae1^–/– mice showed no difference in brain size and gross cortical morphology, including the thickness and the number of neurons in the somatosensory cortex, compared with PV-Cre mice (Supplemental Figure 2).

Figure 1

Spontaneous seizures and premature death in Nae1-deficient mice. (A) Schematic diagram of breeding strategy to generate PV-Nae1^–/– and PV-Cre mice. (B) Specific ablation of Nae1 in PVINs. Hippocampus sections of PV-Cre (top) and PV-Nae1^–/– (bottom) mice were stained with anti-Nae1 (green) and anti-PV (red) antibodies. Nuclei are shown in blue (DAPI staining). Arrows show PV-positive cells, and arrowheads show PV-negative cells. Scale bar: 50 μm. n = 3 experiments. (C) Representative EEG traces of freely moving PV-Cre and PV-Nae1^–/– mice. Scale bar: 20 s, 2 mV. (D) Quantitative analysis of seizure numbers within 8 hours. n = 8 mice for each genotype. PV-Cre (0 ± 0 seizure numbers) vs. PV-Nae1^–/– (2 ± 0.46 seizure numbers). P < 0.001. (E) Survival curves of PV-Cre (n = 28) and PV-Nae1^–/– (n = 26) mice. (F–J) No change in PVIN number in hippocampus of PV-Nae1^–/– mice. (F) Fluorescence images showing the PVINs in the hippocampus. Scale bar: 500 μm. (G) Schematic representation of coronal sections of the hippocampus. Numbers indicate distance from bregma. (H–J) Quantification of PVINs in CA1, CA2/3, and DG regions of the 3 coronal sections. n = 7 slices, 3 mice for each genotype. Slices were stained with anti-NeuN (red) and PV (green) antibodies; nuclei are shown in blue (DAPI staining). Data are represented as mean ± SEM. ***P < 0.001, unpaired 2-tailed Student’s t test.

PV-Nae1^–/– mice exhibited signs of ataxia and seizure in a development-dependent manner. Beginning on P18, they showed limb tremors and imbalanced or uncoordinated walking or stumbling (Supplemental Video 1). None of these phenotypes were observed in control PV-Cre mice at any ages. We assessed ataxia phenotypes by behavioral tests. As shown in Supplemental Figure 3, A–C, in the balance beam test, the latency and distance walked prior to falling from the beam were reduced in PV-Nae1^–/– mice compared with that in control PV-Cre mice (latency: 14.5 ± 6.9 s in mt versus 60.0 ± 0.0 s in control, P < 0.0001; distance walked: 9.1 ± 1.6 cm in mt versus 122.8 ± 13.2 cm in control, P < 0.0001; n = 8 mice per genotype). To assess the motor reflex, mice were tail suspended above a wire lid of a cage, as described previously (ref. 41, see also Methods). Control mice would outstretch forelimbs to reach downward for the cage lid (Supplemental Figure 3D). However, PV-Nae1^–/– mice displayed a curved body posture and seemingly were unable to reach for the lid. The reflex response time was reduced in mt mice compared with control mice (Supplemental Figure 3E; 29.7% ± 4.09% in mt versus 96.8% ± 1.33% in control, P < 0.0001; n = 8 mice per genotype). These results indicate that the Nae1 mutation in PVINs causes ataxia-like phenotypes in mice. PV-Nae1^–/– mice began to display epileptic seizures around P30 (Supplemental Video 2), which were associated with epileptiform discharges by EEG recording (Figure 1C). At P40, 1 seizure onset was observed every 4 hours (Figure 1D), and the frequency and severity of seizures were increased with age. At P47, PV-Nae1^–/– mice began to die, which resulted from a general seizure or sudden unexpected death in epilepsy (Figure 1E). These results demonstrate a critical role of NAE1 of PVIN in regulating brain activity.

A necessary role of Nae1 for PVIN excitability. To investigate the underlying mechanisms, we quantified the number of PVINs in PV-Nae1^–/– mice in the hippocampus, but failed to detect any difference between the mt and control mice (Figure 1, F–J). Similar results were observed in the medial prefrontal cortex (mPFC) and amygdala (Supplemental Figure 4). Next, we recorded spontaneous excitatory postsynaptic currents (sEPSCs) in CA1 PyNs in whole-cell patch configurations without blocking glutamatergic and GABAergic transmission. Interestingly, the frequency of sEPSC in CA1 PyNs was increased without altered amplitude (Figure 2, A–C; sEPSC frequency, 2.64 ± 0.30 Hz in mt versus 1.66 ± 0.32 Hz in control, P < 0.05; sEPSC amplitude 19.17 ± 0.99 pA in mt versus 18.60 ± 0.65 pA in control, P = 0.6372; n = 9 neurons from 3 mice per genotype), indicating increased PyN activity in mt mice. To determine whether this was due to altered function of PVINs or PyNs, we next measured spontaneous inhibitory postsynaptic currents (sIPSCs) of CA1 PyNs in the presence of 6-cyano-7-nitroquinoxaline-2,3-dione (CNQX) and DL-2-amino-5-phosphonopentanoic acid (DL-AP5) (antagonists of AMPA receptors and NMDA receptors, respectively) to block glutamatergic transmission. As shown in Figure 2, D and E, sIPSC frequency was reduced in PV-Nae1^–/– mice compared with that in PV-Cre mice (sIPSC frequency, 6.92 ± 0.93 Hz in mt versus 11.39 ± 1.16 Hz in control, P < 0.01; n = 12 neurons from 4 control mice; n = 11 neurons from 4 mt mice). However, Nae1 mutation had little effect on sIPSC amplitudes (Figure 2F; sIPSC amplitude, 43.4 ± 4.36 pA in mt versus 39.9 ± 2.70 pA in control, P = 0.503). These results suggest that the Nae1 mutation may decrease GABAergic transmission from PVINs onto PyNs in the hippocampus. Next, we recorded sEPSCs when GABAergic transmission was blocked by bicuculline (an antagonist of GABAA receptors), but failed to detect any difference in sEPSC frequency or amplitudes (Figure 2, G–I; sEPSC frequency, 1.49 ± 0.18 Hz in mt versus 1.48 ± 0.16 Hz in control, P = 0.9501; sEPSC amplitude, 17.7 ± 1.00 pA in mt versus 16.5 ± 0.71 pA in control, P = 0.3407; n = 11 neurons from 4 mice per genotype). These observations indicate that the increase in sEPSC is likely caused by reduced GABA release from PVINs.

Figure 2

Reduced GABAergic neurotransmission in PV-Nae1^–/– hippocampus. (A) Representative sEPSC traces of CA1 PyNs without blocking GABAergic and glutamatergic transmission. Left, schematic diagram of recording. Scale bar: 3 s, 40 pA. (B) Increased sEPSC frequency. n = 9 neurons, 3 mice for each genotype. (C) No difference in sEPSC amplitudes. (D) Representative sIPSC traces of CA1 PyNs with blocking glutamatergic transmission (right). Left, schematic diagram of recording. Scale bar: 1 s, 50 pA. (E) Decreased sIPSC frequency. n = 11–12 neurons, 4 mice for each genotype. (F) No difference in sIPSC amplitudes. (G) Representative sEPSC traces of CA1 PyNs with blocking GABAergic transmission (right). Left, schematic diagram of recording. Scale bar: 3 s, 40 pA. (H) No difference in sEPSC frequency. n = 11 neurons, 4 mice for each genotype. (I) No difference in sEPSC amplitudes. Data are represented as mean ± SEM. *P < 0.05; **P < 0.01, unpaired 2-tailed Student’s t test.

To investigate the mechanisms of reduced GABA release in PV-Nae1^–/– mice, we examined the intrinsic excitability of CA1 PVINs by clamping in a whole-cell configuration. APs of PVINs were elicited by injecting depolarizing currents at different intensities in the presence of bicuculline, CNQX, and DL-AP5 to block both inhibitory and excitatory neurotransmission. Intriguingly, the firing frequency of PVINs in response to depolarizing currents was lower in PV-Nae1^–/–-Ai9 mice compared with PV-Ai9 mice (Figure 3, A and B; gene effect, F_1,196 = 34.36, P < 0.0001; interaction effect, F_6,196 = 3.15, P < 0.01; 2-way ANOVA). A similar reduction in PVIN firing frequency was observed in the mPFC and amygdala of PV-Nae1^–/–-Ai9 mice (Supplemental Figure 5). In contrast, there was no difference in the resting membrane potential (RMP), membrane capacitance (C_m), and input resistance (R_in) of PVINs between PV-Ai9 and PV-Nae1^–/–-Ai9 mice (Figure 3, C–E). Importantly, the firing frequency of PyNs in the presence of bicuculline, CNQX, and DL-AP5 was similar between the 2 genotypes (Figure 3, F and G; gene effect, F_1,168 = 0.2578, P = 0.6123; interaction effect, F_6,168 = 0.158, P = 0.9872; 2-way ANOVA). Moreover, no difference was observed in passive membrane properties of PyNs (Figure 3, H–J), including RMP, C_m, and R_in of PyNs between PV-Ai9 and PV-Nae1^–/–-Ai9 mice. However, in the absence of glutamatergic and GABAergic transmission blockers, PyN firing was increased in PV-Nae1^–/–-Ai9 mice compared with controls (Figure 3, K and L; gene effect, F_1,119 = 27.52, P < 0.0001; interaction effect, F_6,119 = 2.568, P < 0.05; 2-way ANOVA). Together, these results indicate that Nae1 deletion in PVINs reduces the intrinsic excitability of PVINs, but not that of PyNs. The resulting reduced GABA release subsequently led to increased PyN firing and glutamatergic transmission. The effect of Nae1 mutation on PVINs was specific because it had little effect on the excitability of cholecystokinin-positive INs (CCKINs) (Supplemental Figure 6, A–C) or the RMP and current-voltage relationship of astrocytes that may regulate neuronal activity (Supplemental Figure 6, D–H).

Figure 3

Impaired PVIN intrinsic excitability in PV-Nae1–deficient mice. (A and B) Impaired PVIN intrinsic excitability of PV-Nae1–deficient mice with blocking glutamatergic and GABAergic transmissions. Schematic diagram of recording (left in A) and representative AP traces (right in A). Quantification of current injection–induced APs in PVINs (B). n = 15 neurons, 4 mice for each genotype. Scale bar: 0.2 s, 30 mV. (C–E) No difference in RMP (C), C_m (D), or R_in (E) in PVINs between the 2 groups. n = 19 neurons, 4 mice for each genotype. (F and G) No change in PyN intrinsic excitability in CA1 of PV-Nae1–deficient mice. Schematic diagram of recording (left in F) and representative AP traces (right in F). Quantification of current injection–induced APs in PyNs (G). n = 12–14 neurons, 4 mice for each genotype. Scale bar: 0.2 s, 40 mV. (H–J) No difference in RMP (H), C_m(I), or R_in (J) in CA1 PyNs between the 2 groups. n = 26 neurons, 4 mice for each genotype. (K and L) Increased PyN network excitability in CA1 of PV-Nae1–deficient mice. Schematic diagram of recording (left in K) and representative AP traces (right in K). Quantification of current injection–induced APs (L) in PyNs. n = 9–10 neurons, 3 mice for each genotype; Scale bar: 0.2 s, 40 mV. Data are represented as mean ± SEM. *P < 0.05; **P < 0.01; ***P < 0.001, unpaired 2-tailed Student’s t test (C, D, E, H, I, and J); 2-way ANOVA (B, G, and L).

Sodium-current deficiency to decrease PVIN excitability in PV-Nae1^–/–-Ai9 mice. The excitability of PVINs is regulated by many proteins, including neural proteoglycan brevican (BCAN), ETS variant transcription factor 1 (ER81), and various ion channels. including hyperpolarization-activated cyclic nucleotide-gated potassium channel 4 (HCN4), voltage-gated potassium channels (KCNA1, KCNC1-3, KCNQ2, and KCNQ3), and the α subunits of voltage-gated sodium channel (SCN1A, SCN3A, and SCN8A, refs. 7, 8, 12, 17–19, 42–45). To investigate how Nae1 mutation regulates PVIN excitability, we analyzed the mRNA levels of these genes by quantitative RT-PCR analyses; no differences were observed between PV-Ai9 and PV-Nae1^–/–-Ai9 mice, suggesting that Nae1 mutation in PVINs did not alter these proteins’ synthesis transcription (Supplemental Figure 7A).

To reveal the mechanisms underlying decreased PVIN excitability, we analyzed the APs elicited in response to individual injection of 400 pA depolarizing current ramp in tdTomato⁺ INs (Figure 4A). APs of these INs had a large and rapid afterhyperpolarization (AHP) phase (Figure 4B), as previously reported for PVINs (46, 47). Compared with control PVINs, rheobase (the minimal current to trigger the first AP) was increased in Nae1 mt PVINs (Figure 4C, rheobase, 193.3 ± 17.7 pA in mt versus 128.9 ± 12.0 pA in control, P < 0.01), consistent with reduced excitability in mt PVINs. To determine which channels may be altered by Nae1 mutation, we measured AP threshold (membrane potential to generate AP) and amplitude, the 2 events dependent on voltage-gated sodium channels (Na_v; refs. 48–51). As shown in Figure 4D, AP thresholds were increased (less negative) in mt PVINs compared with those in controls (threshold, –40.4 ± 1.12 mV in mt versus –44.4 ± 1.15 mV in control, P < 0.05; n = 19 neurons from 4 mice per genotype). AP amplitudes were reduced in PVINs of Nae1 mt mice compared with those in control mice (Figure 4E; amplitude, 49.6 ± 1.88 mV in mt versus 56.3 ± 1.60 mV in control, P < 0.05). These results suggest a deficiency of Na_v in Nae1 mt PVINs. Notice that the effects of Nae1 mutation on sodium channels were specific because the mutation had little effect on AHP amplitudes, which require Ca²⁺-activated voltage-dependent K⁺ channels (Figure 4F; 17.4 ± 0.98 mV in mt versus 16.6 ± 0.93 mV in control, P = 0.5839). Nae1 mutation had no effect on the AP half-widths (Figure 4G; 0.50 ± 0.02 ms in mt versus 0.48 ± 0.02 ms in control, P = 0.3744), suggesting voltage-gated potassium channels and calcium channels might not be affected (51–53). These results indicate that Nae1 mutation increased rheobase and thresholds to generate APs in PVINs, suggesting the deficient Na_v as a potential mechanism (16).

Figure 4

Decreased Na current density and reduced Na_v1.1 level and stability in PV-Nae1–deficient mice. (A and B) Schematic diagram of recording (A) and representative second AP waveform in CA1 PVINs evoked by a 400 pA ramp current injection (B). Scale bar: 2 ms, 10 mV. (C–G) Increased rheobase (C) and AP threshold (D). Decreased AP amplitude (E). No changed AHP amplitude (F) and AP half-width (G). n = 19 neurons, 4 mice for each genotype. (H) Representative traces of voltage-gated sodium currents in CA1 PVINs. Scale bar: 5 ms, 5 pA/pF. (I) Current-voltage curves. n = 9 neurons, 4 mice for each genotype. (J) Reduction of the maximum sodium-current density in PV-Nae1^–/–-Ai9 mice. (K) Comparison of the voltage-dependence activation of sodium channels. Conductance was normalized to the maximal sodium conductance; lines were generated by Boltzmann fitting. Membrane potentials for half-maximal activation and slope factors were as follow: PV-Ai9, -35.10 ± 0.77 mV and 5.51 ± 0.66; PV-Nae1^–/–-Ai9, –33.38 ± 0.87 mV and 5.75 ± 0.76. (L and M) Levels of Na_v1.1 protein were lower in PV-Nae1^–/–-Ai9 mice. (L) Representative blots for hippocampal tissues. (M) Quantification analysis of data. (N and O) Accelerated Na_v1.1 degradation by neddylation inhibition. (N) tsA-201 cells were cotransfected with human Na_v1.1 channel β₁ and β₂ subunits, treated with DMSO, MLN4924, or MG132. (O) Quantification analysis of data. n = 4 experiments. (P) Increased ubiquitinated Na_v1.1 in PV-Nae1^–/–-Ai9 mice. Cerebellar lysates were subjected to immunoprecipitation with anti-Na_v1.1 antibody and immunoblotting with anti-UB (K48) antibody. n = 3 experiments. Data are represented as mean ± SEM. *P < 0.05; **P < 0.01; ***P < 0.001, unpaired 2-tailed Student’s t test (C, D, E, F, G, J, and M); 2-way ANOVA (O).

To further test this idea, we analyzed somatic Na⁺ currents in CA1 PVINs by using a prepulse protocol (ref. 54 and Figure 4H). The Na⁺ currents could be completely abolished by tetrodotoxin, a specific blocker of sodium channels (data not shown). Voltage-clamp step recording showed that the Na_v current density (normalized by cell capacitance) was reduced in PV-Nae1^–/–-Ai9 mice compared with that in PV-Ai9 mice (Figure 4, I and J; max current density 62.3% ± 9.92% in mt versus 100% ± 5.89% in control, P < 0.01; n = 9 neurons from 4 mice per genotype). Nae1 mutation had little effect on voltage-gated calcium currents or hyperpolarization-activated cyclic nucleotide-gated channel currents in PVINs (Supplemental Figure 7, B–F). Notice that the voltage dependence of activation for Na⁺ currents was not changed (Figure 4K), suggesting that mt PVINs have normal Na_v electrophysiological properties. These results suggest that Nae1 mutation reduced Na_v currents without changing Na_v’s properties.

A necessary role of neddylation for Na_v1.1 stability. In the mammalian CNS, Na_v1.1, and Na_v1.6 are abundant in PVINs, whereas Na_v1.2 is mainly expressed in excitatory neurons. Na_v1.3 is transiently expressed in the brain during embryonic development and normally absent in the adult (7, 11, 55). To determine whether Na_v1.1 protein was reduced in PVINs of PV-Nae1^–/–-Ai9 mice, we did WB analysis. As shown in Figure 4, L and M, Na_v1.1, but not Na_v1.6 and Atp1a1 (Na⁺/K⁺-ATPase α 1 subunit), were reduced in the hippocampus of PV-Nae1^–/–-Ai9 mice. Na_v1.1 reduction was also observed in cerebral cortex and cerebellum (Supplemental Figure 8), brain regions where PVINs are abundant. These results support the idea that Na_v1.1 reduction may be a potential mechanism for decreased excitability of PVINs. This notion was supported by increased AP failure rate in PVINs of PV-Nae1^–/–-Ai9 mice compared with PV-Ai9 mice (Supplemental Figure 9). These electrophysiologic deficits, as well as ataxia and seizure phenotypes (Figure 1 and Supplemental Figure 3), were similar to those of Na_v1.1 mt mice (7, 8, 12, 16, 41). Together, these results suggest Na_v1.1 deficiency as a pathology mechanism.

Next, we investigated the molecular mechanisms underlying Na_v1.1 protein reduction in PV-Nae1^–/–-Ai9 mice. As shown in Supplemental Figure 6A, Na_v1.1 mRNA levels were similar between the genotypes. Because neddylation is known to regulate protein stability (56–58), we wondered whether Na_v1.1 stability requires neddylation. To this end, hNa_v1.1 and 2 β subunits were transfected into human tsA-201 cells, SV40-transformed HEK293 cell that are able to express ion channels on the membrane (59). Forty-eight hours after transfection, cells were incubated with 50 μM cycloheximide (CHX) to inhibit protein synthesis and with MLN-4924, a potent and selective inhibitor of NAE (60, 61). Interestingly, Na_v1.1 degradation was increased in MLN-4928–treated cells compared with nontreated cells (Figure 4, N and O), suggesting a critical role of neddylation for Na_v1.1 stability. In accordance with this, K48-linked ubiquitination was increased in PV-Nae1^–/–-Ai9 mice compared with PV-Ai9 mice (Figure 4P). These results suggest that neddylation may be critical to the stability of Na_v1.1.

Neddylation-dependent surface expression of Na_v1.1. To determine whether Na_v1.1 is subjected to modification by neddylation, we transfected tsA-201 cells with Na_v1.1-Myc together with HA-Nedd8. Na_v1.1 was immunoprecipitated by anti-Myc antibody and probed with anti-HA antibody. Interestingly, HA signal was detected in precipitated Na_v1.1; importantly, the signal was reduced in cells that were incubated with the NAE inhibitor MLN-4924 (Figure 5A). These results suggest that the channel may be neddylated in tsA-201 cells. To determine whether Na_v1.1 neddylation occurs in vivo, Na_v1.1 was precipitated with anti-Na_v1.1 antibody from cerebellar homogenates (where PV⁺ cells are abundant) and probed with anti-Nedd8 antibody, as previously described (36, 62). Neddylated Na_v1.1 was reduced in PV-Nae1–deficient mice compared with control mice (Figure 5B). Together, these results demonstrate that Na_v1.1 is neddylated in cultured cells and in the brain.

Figure 5

Reduced surface expression of K1936E mt Na_v1.1. (A) Inhibiting Na_v1.1 neddylation by MLN4924. tsA-201 cells were transfected with Na_v1.1-Myc and HA-Nedd8 plasmids and treated with or without MLN-4924. n = 3 experiments. (B) Decreased Na_v1.1 neddylation in PV-Nae1^–/–-Ai9 mice. Neddylated Na_v1.1 precipitated with anti-Na_v1.1 antibody and probed with anti-NEDD8 antibody. n = 3 experiments. (C) Accelerated degradation of K1936E. tsA-201 cells were cotransfected with Na_v1.1-WT, K245N or K1936E with β₁ and β₂ subunits, treated with CHX for indicated times, and subjected to immunoblotting. n = 3 experiments. (D) Quantification analysis of data in C. (E–G) Reduced total and surface K1936E Na_v1.1 in tsA-201 cells. Representative blots for tsA-201 cells (E), quantitative data of total (F) and surface (G) levels. n = 3 experiments. (H and I) Reduced neddylation at K1936E. (H) Na_v1.1-WT, K245N, and K1936E were transiently transfected in tsA-201 cells together with HA-Nedd8. (I) Quantification analysis of data. n = 3 experiments. Data are represented as mean ± SEM. *P < 0.05; **P < 0.01; ***P < 0.001, 1-way ANOVA (F, G, and I); 2-way ANOVA (D).

To further investigate how neddylation regulates Na_v1.1, we sought to identify the lysine residues in Na_v1.1 that may be subjected to neddylation. We screened for potential pathogenic lysine mutations in databases of epilepsy patients and the ClinVar in the US National Library of Medicine. Sixteen lysine missense mutations were reported in patients with epilepsy (63–69); in particular, K1936E (where lysine1936 was mutated to glutamic acid) was identified in a screen of 70 genes in 8565 patients, including 8354 epileptic patients (68) (see also Human SCN1A gene database; http://scn1a.caae.org.cn/index.php). Because K1936 is localized in the cytoplasmic region of Na_v1.1, we determined whether its mutation alters the stability of the Na_v1.1 protein. WT-Na_v1.1 or K1936E, along with 2 β subunits, was transfected into tsA-201 cells, and their stability was studied in the presence of CHX. Compared with WT-Na_v1.1, K1936E was degraded faster (Figure 5, C and D). As a control, the stability was not changed for K245N, another lysine mutation also located in the cytoplasmic domain. In agreement with this, surface and total levels of K1936E, but not K245N, were reduced (Figure 5, E–G). Concomitantly, Na_v1.1 neddylation was reduced in cells expressing K1936E, but not K245N, indicating that Lys1936 is a primary neddylation site (Figure 5, H and I). These results suggest that neddylation at Lys1936 is important in maintaining the stability of Na_v1.1 and that K1936E mt is less stable, identifying a potential pathological mechanism of this mutation.

To investigate whether the functional properties of Na_v1.1 channels were changed by these mutations, tsA-201 cells were transfected with WT-Na_v1.1, K245N, or K1936E, together with β₁ and β₂ subunits and GFP (70), and were recorded by patch clamp in whole-cell configuration. The current-voltage curves showed that the current density was reduced in cells expressing K1936E, but not K245N, compared with WT (Figure 6, A and B). Notice that the coprecipitation of mt Na_v1.1 and β₁ or β₂ subunits was comparable to that of WT-Na_v1.1, suggesting that the K1936E mutation had no impact on Na_v1.1 interaction with β₁ or β₂ subunits (Supplemental Figure 10). Next, we characterized the kinetics of the sodium-channel activation and inactivation. As shown in Figure 6, C and D, the curves of activation and inactivation were similar between WT-Na_v1.1 and K1936E, suggesting that the mutation did not change the voltage-dependent activation and inactivation of sodium currents, respectively. In addition, K245N showed a depolarizing shift of steady-state fast inactivation. Together, these results demonstrate that the K1936E mutation reduces Na_v1.1 levels without altering the channel property.

Figure 6

Reduced Na_v current density by the K1936E mutation. (A) Representative whole-cell sodium currents. tsA-201 cells were injected with depolarizing steps from –80 to +50 mV, with a holding potential of –120 mV. Scale bar: 2 ms, 50 pA/pF. (B) Current-voltage relationships. Whole-cell sodium currents were normalized to cell capacitance. n = 10–11 cells for each group. Sodium-current density is significantly decreased in K1936E group compared with WT group. (C) Comparison of the voltage-dependence activation of sodium channels. Conductance was normalized to the maximal sodium conductance between –80 and +20 mV; lines represent mean Boltzmann fits. n = 10–11 cells for each group. The membrane potentials for half-maximal activation and slope factors were as follows: WT, –29.3 ± 0.98 mV and 8.8 ± 0.9; K245N, –29.0 ± 1.05 mV and 9.3 ± 0.97; K1936E, –28.3 ± 1.04 mV and 9.0 ± 0.95. (D) Depolarizing shift (+6.8 mV) of steady-state fast inactivation of Na_v1.1 with K245N mutation. Currents were normalized to the peak current amplitude; lines represent mean Boltzmann fits. n = 10–11 cells for each group. The membrane potentials for half-maximal inactivation and slope factors were as follows: WT, –52.7 ± 0.60 mV and 10.1 ± 0.53; K245N, –45.9 ± 0.66 mV and 10.5 ± 0.59; K1936E, –54.1 ± 0.72 mV and 10.8 ± 0.64. Data are represented as mean ± SEM.

Altered synaptic functions revealed by proteomic analysis. Sodium-channel levels and function are regulated by a variety of proteins or pathways. For example, increased intracellular calcium concentrations reduced sodium-current density (71). Sodium-channel surface expression and stability require auxiliary β subunits, such as SCN1B and SCN2B (72, 73). Fibroblast growth factor 14 (FGF14) interacts with and inhibits sodium channels (74, 75), whereas loss of FGF14 reduced Na_v1.6 expression (76). Finally, neural precursor cell expressed, developmentally downregulated 4 like (NEDD4L), a ubiquitin E3 ligase, decreases the sodium-channel density at cell surface in Xenopus oocytes and heart cells (77–79). Receptor for activated C kinase 1 (RACK1) inhibits mRNA expression of Na_v1.1 by binding to a silencer downstream of the SCN1A promoter (80). As shown in Supplemental Figure 11, mRNA levels of these proteins were similar between PV-Nae1^–/–-Ai9 mice and PV-Ai9 mice.

To fully understand the impact of the PVIN-specific Nae1 mutation on brain function, we compared the proteomes between PV-Nae1^–/–-Ai9 mice and PV-Ai9 mice (Figure 7A). Out of a total of 5167 proteins identified (Supplemental Table 1), 169 and 279 proteins were down- and upregulated, respectively, in PV-Nae1^–/–-Ai9 mice compared with PV-Ai9 mice (P < 0.05; Figure 7B). As expected, NAE1 was reduced in mt brain samples (Figure 7B), in support of our hypothesis. Gene Ontology (GO) analysis of downregulated proteins implicated a variety of cellular functions (Supplemental Table 2). Interestingly, 9 of the top 20 relate to neural development, neurotransmission, and synaptic plasticity, including dendrite development, synapse organization, synaptic vesicle exocytosis, and glutamatergic transmission (Figure 7C and Supplemental Table 2), in accordance with a previous report (37). In agreement, Kyoto Encyclopedia of Genes and Genomes (KEGG) analysis highlighted glutamatergic and dopaminergic synapses, synaptic vesicle cycle, endocannabinoid signaling, and alcoholism as well as spliceosome (Supplemental Figure 12A and Supplemental Table 3; P < 0.05). On the other hand, for upregulated proteins, GO analysis pointed to ubiquitin-regulated catabolism and protein folding and localization (Figure 7D and Supplemental Table 4), whereas KEGG analysis highlighted metabolic pathways, including propanoate, pyruvate and carbon metabolism, tRNA synthesis, phagocytosis, and ubiquitin-mediated proteolysis (Supplemental Figure 12B and Supplemental Table 5, P < 0.05). These results suggest that Nae1 is required not only for maintaining the excitability and GABA release of PVINs, but also for a plethora of cellular functions, such as RNA splicing, metabolism, protein processing and stability, and phagocytosis.

Figure 7

Proteomic analysis revealed reduced Na_v1.1 and compromised synaptic function and metabolic pathways in PV-Nae1–deficient mice. (A) Schematic diagram of quantitative proteomic analysis in PV-Ai9 and PV-Nae1^–/–-Ai9 mice. n = 3 mice per genotype. (B) Volcano plots of differentially expressed proteins in PV-Nae1^–/–-Ai9 mice over PV-Ai9 mice. Green and red dots indicate significantly down- and upregulated proteins, respectively. (C) Top 20 significantly enriched GO terms of downregulated proteins. (D) Top 20 significantly enriched GO terms of upregulated proteins. (E) Heatmap of ion channel proteins. Green, downregulated ion channels. (F) Heatmap of proteins associated with epilepsy. Green and red colors indicate down- and upregulated proteins, respectively. (G) Twelve ion channels were epilepsy-associated genes, but only Na_v1.1 was significantly reduced by Nae1 mutation.

Considering compromised intrinsic excitability of PVINs in Nae1 mt mice, we wondered whether the levels of other ion channels are altered. Indeed, among the identified proteins were 53 ion channels (Figure 7E and Supplemental Table 1). However, only SCN1A and voltage-gated calcium channel auxiliary subunit γ 7 (CACNG7) were significantly changed (in fact reduced) in PV-Nae1^–/–-Ai9 samples compared with control PV-Ai9 samples (Figure 7E; P < 0.05). CACNG7 is a type II transmembrane AMPA receptor regulatory protein (TARP) that regulates trafficking and channel gating of the AMPA receptors (81); however, CACNG7 mt mice were apparently normal in locomotion and excitatory transmission (81, 82), suggesting that CACNG7 reduction may not contribute to epileptic behavior in Nae1 mt mice. On the other hand, recent studies have identified 29 risk genes for epilepsy (20–22, 83–87) (Figure 7F). Intriguingly, in addition to Na_v1.1/SCN1A, GABAA receptor δ subunit (GABRD, a subunit of extrasynaptic GABAA receptor) and proline-rich transmembrane protein 2 (PRRT2, a synaptic protein implicated in synaptic formation and maintenance) were also reduced (Figure 7F). However, IN-specific deletion of GABRD increased IN excitability and thus decreased PyN excitability and seizure susceptibility (88). PRRT2 mt increased intrinsic excitability of PyNs, and PRRT2 expression decreased sodium currents by Na_v1.2/Na_v1.6, but not Na_v1.1 in HEK293 cells (89). Therefore, the reduced excitability of PVINs by Nae1 deficiency is unlikely to be due to reduced levels of GABRD or PRRT2. K⁺/Cl^– cotransporter 2 (KCC2, encoded by Slc12a5) is a potassium-chloride cotransporter for maintaining the intracellular concentration of chloride ions (90). Its deficiency has been implicated in epileptogenesis (91, 92). As shown in Figure 7F, KCC2 levels were increased in Nae1-deficient mice compared with those in control samples, suggesting that the cause of epilepsy by Nae1 mutation could be complex and may involve KCC2 in addition to reduced levels of Na_v1.1 (see Discussion).

Note that of 53 ion channels and 29 epilepsy-associated genes, 12 were identified in both groups (Figure 7G). Only Na_v1.1 was reduced in PV-Nae1^–/–-Ai9 compared with control samples. The results from proteomic analysis provide further evidence for a role of neddylation for Na_v1.1 stability.

Restoring the excitability of Nae1 mt PVINs by expressing Na_v1.1. Our hypothesis was that Nae1 deficiency reduces Na_v1.1 and thus the excitability of PVINs, which causes epilepsy. To further test this, we determined whether expressing Na_v1.1 in Nae1 mt PVINs ameliorates the impact of Nae1 mutation on the excitability. LSL-GFP and hNa_v1.1-Myc plasmids were injected into the cortex of PV-Nae1^–/–-Ai9 mice (Figure 8A and ref. 93). Expression of GFP was dependent on Cre. As shown in Figure 8B, GFP and hNa_v1.1-Myc were detected specifically in PVINs that were labeled by tdTomato in PV-Nae1^–/–-Ai9 mice. First, we determined whether hNa_v1.1 expression increased somatic Na⁺ currents in GFP⁺ INs. Voltage-clamp step recording showed that the Na_v current density was lower in LSL-GFP–injected PV-Nae1^–/–-Ai9 mice compared with LSL-GFP–injected PV-Ai9 mice; however, the Na_v current density in mt mice injected with LSL-GFP and hNa_v1.1-Myc was increased to a level comparable to that of control mice (Figure 8, C and D, max current density 100% ± 8.92% in wt + LSL-GFP versus 64.4% ± 9.30% in mt + LSL-GFP, P < 0.05; 116.4% ± 9.66% in mt + LSL-GFP & hNa_v1.1-Myc versus 64.4% ± 9.30% in mt + LSL-GFP, P < 0.01; n = 8–10 neurons from 4 mice per group). Next, we analyzed the intrinsic excitability, rheobase, and single AP properties of GFP⁺ INs in PV-Nae1^–/–-Ai9 mice. Compared with PVINs expressing LSL-GFP, neurons expressing hNa_v1.1-Myc generated more APs in response to injected currents, indicative of improved excitability (Figure 8, E and F; expression effect, F_1,133 = 61.8, P < 0.0001; interaction effect, F_6,133 = 4.71, P < 0.001; 2-way ANOVA). Na_v1.1 expression also decreased AP rheobases (Figure 8G; 116.2 ± 11.2 pA in hNa_v1.1 versus 182.7 ± 23.1 pA in control, P < 0.05; n = 14 neurons from 5 mice per genotype) and thresholds (Figure 8H; –46.1 ± 1.53 mV in hNa_v1.1 versus –40.1 ± 1.29 mV in control, P < 0.01)and increased AP amplitudes (Figure 8I; 57.5 ± 1.11 mV in hNa_v1.1 versus 51.2 ± 1.30 mV in control, P < 0.01), but had no effect on AHP amplitudes and AP half-widths (Figure 8, J and K; AHP amplitude, 18.5 ± 1.14 mV in hNa_v1.1 versus 17.8 ± 1.11 mV in control, P = 0.6348; AP half-width, 0.47 ± 0.02 ms in hNa_v1.1 versus 0.45 ± 0.02 ms in control, P = 0.5489). These findings demonstrate that restoring hNa_v1.1 in neddylation-deficient PVINs was sufficient to recover the excitability, suggesting that Na_v1.1 deficiency is a major mechanism of Nae1 mutation in regulating PVIN excitability. Taken together, these results indicate that Na_v1.1 requires neddylation for stable expression in PVINs and thus their excitability.

Figure 8

Restoration of Nae1 mt PVIN excitability by expressing Na_v1.1. (A) A diagram of in vivo electroporation. (B) Expression of hNa_v1.1-Myc and GFP in PVINs. GFP (green) signaling colocalized with anti-Myc (blue). White arrows indicate PVINs. n = 3 experiments. Scale bars: 40 μm (top row); 10 μm (bottom row). (C) Current-voltage curves of sodium-current density in GFP⁺ neurons. n = 8–10 neurons, 5 mice for each genotype. (D) Rescue of sodium-current density by Na_v1.1 expression in PV-Nae1–deficient mice. (E and F) Increased excitability of mt PVINs after expressing Na_v1.1. Representative AP traces and quantification of current injection-induced APs in GFP⁺ IN. n = 10–11 neurons, 5 mice for each group. Scale bar: 0.2 s, 30 mV. (G–K) Rescue of AP deficits of PVINs by Na_v1.1 expression in PV-Nae1 deficient mice. Decreased rheobase (G) and AP threshold (H). Increased AP amplitude (I). No effect on AHP amplitude (J) and AP half-width (K). n = 14 neurons, 5 mice for each group. Data are represented as mean ± SEM. *P < 0.05; **P < 0.01, 1-way ANOVA (D); 2-way ANOVA (F); unpaired 2-tailed Student’s t test (G, H, I, J, and K).

In the present study, we uncovered a previously unknown mechanism for regulating the intrinsic excitability of PVINs. Loss of the obligatory subunit of neddylation E1, NAE1, reduced the intrinsic excitability of PVINs, but had little effect on their RMP, C_m, and R_in. As a result, AP thresholds were increased and amplitudes were reduced. Ensuing reduced GABA transmission and E-I imbalance resulted in spontaneous epileptic seizures. These results reveal a critical role of neddylation in regulating the excitability of INs. Further mechanistic studies identified Na_v1.1 as a major target for Nae1 mutation because voltage-clamp step recording indicated that Nae1 mutation reduced the Na-current density without changing its voltage dependence. In accordance with this, WB revealed lower levels of the Na_v1.1 protein in Nae1 mt mice or cells where neddylation was pharmacologically inhibited. This idea was supported by proteomic analysis, which also revealed abnormalities in synapses and metabolic pathways (see below). Finally, Na_v1.1 expression restored the excitability in PVINs in Nae1 mt mice. These results demonstrate that neddylation is necessary for the stability of Na_v1.1 in PVINs, revealing what we believe is a new regulatory mechanism for the excitability of INs.

Implicated in gene transcription and proteolysis, neddylation regulates cell division, differentiation, and apoptosis (33–35). Nae1 mutation in developing brain (via Emx1-Cre) impairs cortical lamination and in adult PyNs (via CaMKII-CreER) reduces spine numbers, suppresses evoked excitatory postsynaptic potential, and impairs learning and memory in mt mice (37, 94). Rapsyn, a classic adaptor protein, possesses E3 ligase activity that regulates acetylcholine receptor clustering and neuromuscular junction formation (36, 95). Here, we show that PVIN-specific mutation of Nae1 causes epilepsy and reduces the intrinsic excitability of PVINs, but has no apparent effects on that of PyNs (Figure 1C and Figure 3). No significant changes in mRNA and protein levels were observed for proteins known to regulate the excitability of PVINs, including HCN4 and voltage-gated potassium channels (Figure 7E and Supplemental Figure 7A). Among the α subunits of voltage-gated sodium channels that are critical to PVIN excitability, only Na_v1.1 was reduced by WB analysis and nonbiased proteomic analysis (Figure 4M and Figure 7E). In agreement, first, AP thresholds and amplitudes, both requiring sodium channels, were increased and decreased, respectively (Figure 4, D and E). Second, the Na_v current density was reduced in PV-Nae1^–/–-Ai9 mice compared with control mice (Figure 4, I and J). However, the Nae1 mutation had little effect on voltage-gated calcium channels and hyperpolarization-activated cyclic nucleotide-gated channels in PVINs (Supplemental Figure 7, B–F). Third, epileptic and electrophysiological phenotypes were similar between PV-Nae1^–/–-Ai9 mice and Na_v1.1 heterozygous mt mice (7, 12, 16, 23, 41). Finally, in vivo expression of Na_v1.1 increased the intrinsic excitability of PVINs, reduced AP thresholds, and increased AP amplitudes (Figure 8, H and I). A parsimonious interpretation of these results is that neddylation is critical to PVIN intrinsic excitability, acting by maintaining Na_v1.1 stability. In support of this notion, proteomic analysis and RT-PCR failed to detect changes in proteins that regulate sodium-channel levels and functions, including auxiliary β subunits of sodium channels, FGF14, NEDD4L, and RACK1 (Supplemental Figure 11).

Mutations or loss of function in Na_v1.1 causes a spectrum of epileptic syndromes such as SMEI and GEFS⁺. T875M and R1648H were the first missense mutations identified in patients with GEFS⁺ (20), and L986F was initially identified in SEMI (21). Since then, more dysfunctional mutations have been identified and have appeared to have different pathological mechanisms. For example, T875M and W1204R mutations exhibited depolarizing and hyperpolarizing shifts of channel activation, respectively (70). The R1657C mutation reduced current density, depolarizing shift of channel activation and fast recovery of slow inactivation (96). Besides the biophysical properties, mutations R1648C and G1674R may impair intracellular trafficking and reduce surface expression (28). The D1866Y mutation was reported to disrupt the interaction between Na_v1.1 and the β₁ subunit, altering the voltage dependence and kinetics (97). Our study shows that the K1936E mutation could reduce Na_v1.1 surface expression by impairing its neddylation. In agreement with this, K48-linked ubiquitination was increased in PV-Nae1^–/–-Ai9 mice compared with PV-Ai9 mice, suggesting that neddylation could increase Na_v1.1’s stability by inhibiting ubiquitination. Our study thus provides what we believe is a previously unknown pathological mechanism of Na_v1.1 mutations. The K245N mutation, located within the cytosolic S4-S5 linker of domain I in Na_v1.1 (65), may act in a pathogenic way by decreasing fast inactivation of sodium channels without altering neddylation.

The E-I balance is not only controlled by the intrinsic excitability of PVINs, but also that of CCKINs, a major source of inhibitory signals that has been implicated in epilepsy (98). On the other hand, astrocytes could modulate the neuronal activity by altering extracellular concentrations of ATP, potassium, and glutamate and thus epileptogenesis (99–102). However, Nae1 mutation in PVINs had little effect on CCKIN intrinsic excitability and astrocytic electrophysiological properties (Supplemental Figure 5). Proteomic analysis indicates that, of the currently known 29 epilepsy-associated risk genes, in addition to Na_v1.1, GABRD and PRRT2 were reduced (Figure 7F). However, IN-specific deletion GABRD increased IN excitability, as described earlier (88). PRRT2 mt mice displayed increased intrinsic excitability of PyNs (89), which was not observed in Nae1 mt mice. Moreover, PRRT2 coexpression had little effect on Na_v1.1 in HEK293 cells (89). However, global GABRD knockout decreased inhibition onto granule cells and thus increased seizure susceptibility (103); PRRT2 reduction in PyNs was shown to increase the excitability of PyNs by increasing sodium-channel currents (89) and is thus likely to cause epilepsy. These suggest that the pathological mechanisms of epilepsy may be complex and may involve reduced GABRD and PRRT2 in other cells. Because Nae1 was ablated specifically in PVINs, these effects may be secondary. In addition, proteomic analysis also revealed an increase in KCC2 in PV-Nae1^–/–-Ai9 mice compared with controls (Figure 7F). Low levels of KCC2 enable GABA to be excitatory in PyNs, whereas increased KCC2 in PyNs renders GABA to be inhibitory (104). These observations suggest that KCC2 increase may result from a compensatory mechanism.

Interestingly, GO and KEGG analyses of Nae1 mutation–downregulated proteins highlighted functional processes, such as dendrite development, synapse organization, synaptic vesicle exocytosis, glutamatergic transmission, dopaminergic synapses, endocannabinoid signaling, and alcoholism as well as spliceosome (Figure 7C and Supplemental Figure 12A, and Supplemental Tables 2 and 3). On the other hand, analysis of the upregulated proteins suggested changes in ubiquitin-regulated catabolism, protein folding and localization, and metabolic pathways, including propanoate, pyruvate and carbon metabolism, tRNA synthesis, phagocytosis, and ubiquitin-mediated proteolysis (Figure 7D, Supplemental Figure 12B, and Supplemental Tables 4 and 5). Determining whether these processes contribute to epileptogenesis warrants future studies. Our study identifies a regulatory mechanism of Na_v1.1 stability in PVINs by neddylation and demonstrates that deficient neddylation reduces Na_v1.1 levels, which reduces the excitability of PVINs and thus epilepsy. Increasing Scn1a expression could elevate the excitability of Scn1a-deficient PVINs and attenuate epileptic phenotypes in Dravet syndrome mouse models (105). Selective activators of Na_v1.1 also benefit seizure (106) and other Na_v1.1-associated deficits (107). Further studies are necessary to explore whether modification of neddylation represents a new therapeutic strategy for epilepsy.

Detailed methods are provided in Supplemental Methods.

Statistics. Data were analyzed with GraphPad Prism software. For analysis between 2 groups, 2-tailed unpaired Student’s t test was used. For analysis of multiple groups at single time points, 1-way ANOVA was used, followed by Tukey-Kramer post hoc multiple comparisons test. For analysis of multiple groups at multiple time points, 2-way ANOVA was used, followed by Bonferroni’s post hoc multiple comparisons test. Data were expressed as mean ± SEM. P < 0.05 was considered statistically significant.

Study approval. These studies were approved by the Institutional Animal Care and Use Committee of Augusta University and Case Western Reserve University.