LGI1 acts presynaptically to regulate excitatory synaptic transmission during early postnatal development

Abstract

The secreted leucine-rich glioma inactivated 1 (LGI1) protein is an important actor for human seizures of both genetic and autoimmune etiology: mutations in LGI1 cause inherited temporal lobe epilepsy, while LGI1 is involved in antibody-mediated encephalitis. Remarkably, Lgi1-deficient (Lgi1(-/-)) mice recapitulate the epileptic disorder and display early-onset spontaneous seizures. To understand how Lgi1-deficiency leads to seizures during postnatal development, we here investigated the early functional and structural defects occurring before seizure onset in Lgi1(-/-) mice. We found an increased excitatory synaptic transmission in hippocampal slices from Lgi1(-/-) mice. No structural alteration in the morphology of pyramidal cell dendrites and synapses was observed at this stage, indicating that Lgi1-deficiency is unlikely to trigger early developmental abnormalities. Consistent with the presynaptic subcellular localization of the protein, Lgi1-deficiency caused presynaptic defects, with no alteration in postsynaptic AMPA receptor activity in Lgi1-/- pyramidal cells before seizure onset. Presynaptic dysfunction led to increased synaptic glutamate levels, which were associated with hyperexcitable neuronal networks. Altogether, these data show that Lgi1 acts presynaptically as a negative modulator of excitatory synaptic transmission during early postnatal development. We therefore here reveal that increased presynaptic glutamate release is a key early event resulting from Lgi1-deficiency, which likely contributes to epileptogenesis.

Figure 1. Hippocampal glutamatergic synaptic transmission is increased in Lgi1^−/− mice.

Cumulative distributions and means of mEPSC amplitude (b,d) and inter-event interval (c,e), with sample traces above (a), in CA1 pyramidal cells from Lgi1^+/+ (n = 18) and Lgi1^−/− mice (n = 20) at P8. Scale bar: 20 pA, 500 ms. mEPSC amplitude cumulative distribution and mean, but not inter-event interval, differ significantly between Lgi1^−/− and Lgi1^+/+ mice (p < 0.05, Kolmogorov-Smirnov and t-test, respectively).

Figure 2. Dendritic arborization and excitatory synapses are unaltered in hippocampal pyramidal cells from Lgi1^−/− mice.

(a) Representative dendritic morphology of neurobiotin-filled CA1 pyramidal cells from Lgi1^+/+ and Lgi1^−/− mice at P9. Scale bar: 50 μm. (b) Sholl analysis reveals similar apical dendritic length (p > 0.05, two-way ANOVA) and number of apical dendritic nodes (p > 0.05, Mann-Whitney) between Lgi1^+/+ (n = 9) and Lgi1^−/− mice (n = 14). (c) Electron micrographs of asymmetrical synapses in CA1 stratum radiatum from Lgi1^+/+ and Lgi1^−/− mice at P8. Scale bar: 200 nm. (d) No change was found in asymmetrical synapse density (n = 2900 μm² from 9 Lgi1^+/+ and 6 Lgi1^−/− mice), PSD length (n = 40 synapses from 9 Lgi1^+/+ and 6 Lgi1^−/− mice) and number of docked vesicles (n = 20 synapses from 9 Lgi1^+/+ and 6 Lgi1^−/− mice) (p > 0.05, Mann-Whitney). PSD, postsynaptic density; AZ, active zone.

Figure 3. Postsynaptic AMPAR density is unchanged in Lgi1^−/− mice.

(a) Similar ratios of AMPA to NMDA eEPSCs in CA1 pyramidal cells from P8 Lgi1^+/+ (n = 6) and Lgi1^−/− mice (n = 7; p > 0.05, t-test), with sample traces above. Scale bars: 20 ms, 50 pA. (b) Top: Representative Western blot and quantification of GluR1 and actin bands showing similar GluR1 protein levels in the hippocampus of P8 Lgi1^+/+ (n = 4) and Lgi1^−/− mice (n = 5; p > 0.05, t-test). Bottom: Representative Western blot and quantification of GluR2/3 and actin bands showing similar GluR2/3 protein levels in the hippocampus of P8 Lgi1^+/+ (n = 3) and Lgi1^−/− mice (n = 3; p > 0.05, t-test). (c) Representative immunohistochemical staining for GluR1 (top) and extracellular GluR2 (bottom) subunits in hippocampal sections of Lgi1^+/+ (n = 3) and Lgi1^−/− (n = 3) mice showing similar expression of the AMPAR subunits at P8. Scale bar: 400 μm.

Figure 4. Presynaptic alterations lead to increased synaptic glutamate levels in Lgi1^−/− mice.

(a) Hippocampal electron micrographs from an Lgi1^+/+ mouse showing the subcellular localization of Lgi1, immunostained with patient CSF, in presynaptic elements (arrow) and thin neurites (arrowhead). Non-reactive synaptic contacts are labeled with asterisks. Scale bars: 200 nm. (b) Coefficient of variation (CV) analysis of AMPAR eEPSC amplitude reveals an increase in CV⁻² in CA1 pyramidal cells from P8 Lgi1^−/− (n = 23) compared to Lgi1^+/+ mice (n = 22; p < 0.05, t-test). (c) Synaptic glutamate levels, revealed by γ-DGG (500 μM) inhibition of eEPSC amplitude, are enhanced in P8 Lgi1^−/− (n = 6) compared to Lgi1^+/+ mice (n = 7; p < 0.05, t-test).

Figure 5. Lgi1^−/− mice display glutamatergically-driven spontaneous interictal-like discharges.

(a) Scheme of a hippocampal slice illustrating the arrangement of electrodes for field potential (FP) and whole-cell patch clamp dual recordings. (b) Representative trace of CA1 FP (top) and simultaneous whole-cell patch-clamp recording (bottom) of an Lgi1^−/− CA1 pyramidal cell (resting potential, −70 mV), showing spontaneous interictal-like activities at P9 in normal ACSF. Scale bars: FP: 200 μV, 5 s; whole-cell recording: 20 mV, 5 s. Right: Expanded traces showing an interictal-like network activity (top) that coincides with an intracellular depolarizing plateau potential with burst firing (bottom). Scale bars: FP: 100 μV, 1 s; whole-cell recording: 20 mV, 1 s. Quantification of (c) interictal activity and (d) depolarizing plateau potential in Lgi1^−/− CA1 pyramidal cells (n = 14). (e) Representative traces of CA1 FP (top) and simultaneous pyramidal cell intracellular recordings (bottom) (resting potential, −70 mV) in an Lgi1^−/− slice before (left) and after application (right) of glutamate receptor antagonists NBQX (10 μM) and APV (40 μM), which entirely block interictal-like discharges. Scale bars: FP: 100 μV, 10 s; whole-cell recording: 50 mV, 10 s. AP, action potential.